Programming And Software Engineering

(Faunus/89914) OH YEAH THE LINK HELPS

Faunus@rss.iscabbs.com — Wed, 22 Feb 2012 11:47:00 -0600


OH YEAH THE LINK HELPS

https://stripe.com/blog/capture-the-flag

(Faunus/89913) Y'all might find this fun. I got to level 3 pretty easily, with...

Faunus@rss.iscabbs.com — Wed, 22 Feb 2012 11:09:00 -0600


Y'all might find this fun.  I got to level 3 pretty easily, with the help of a
co-worker who happened to have the right browser extension handy when I didn't.

Level 4 is going to take "actually knowing some C" so I'm not sure how I'll do
with it.

(Feoh/89912) Best Bug Tracking System Ad Ever!

Feoh@rss.iscabbs.com — Wed, 22 Feb 2012 10:36:00 -0600

Best Bug Tracking System Ad Ever!
http://www.youtube.com/watch?v=hMRmOIGaFnw#!

(Feoh/89911) John Public> Exactly, this was a tangled mess where folks made a...

Feoh@rss.iscabbs.com — Wed, 22 Feb 2012 10:18:00 -0600

John Public> Exactly, this was a tangled mess where folks made a series of
disastrous merges.

(John Public/89910) I've been using qgit, but only for viewing the branch/merge hi...

John Public@rss.iscabbs.com — Wed, 22 Feb 2012 10:00:00 -0600

I've been using qgit, but *only* for viewing the branch/merge history to check
whether not what I think happened really happened. I don't use the git guis
for my own development, but I can see how a gui would be indispensable for
making sense of other people's repos.

(Darkhaven/89909) per my question #89895, i got an answer / figured it out:

Darkhaven@rss.iscabbs.com — Wed, 22 Feb 2012 09:46:00 -0600

per my question #89895, i got an answer / figured it out:
http://stackoverflow.com/questions/9370114/jquery-ui-autocomplete-click-event-h
andler

(The After Party/89908) Eclipse SVN Commit Silent Fail

The After Party@rss.iscabbs.com — Wed, 22 Feb 2012 06:02:00 -0600

Eclipse SVN Commit Silent Fail

  I made a change, hit commit and left.
  The next day I come to find out my commit did not work.
  Eclipse silently failed, because there was a change/merge issue
  (my copy wasn't current). Any way to get Eclipse to notify me of this
  so I don't waste time thinking something is committed and its not?

(Feoh/89907) Mandel> Indeed. I think if you're using an SCM you can install ...

Feoh@rss.iscabbs.com — Wed, 22 Feb 2012 04:19:00 -0600

Mandel> Indeed.  I think if you're using an SCM you can install an app outside
the box :) I DO think it's unfortunate that the sandboxing rules don't have
some sort of escape clause - e.g. "The software you are installing can change
random files in your filesystem in the directories you specify.  Capiche?"

(Mandel/89906) Feoh>> Those are the same guys who had a blog post recently not...

Mandel@rss.iscabbs.com — Tue, 21 Feb 2012 17:00:00 -0600

Feoh>>  Those are the same guys who had a blog post recently noting how Apple's
new sandbox rules for the App Store will not work with their software.  Maybe
most programmer don't care to much about the App Store though.

(John Public/89905) Kena> Use the -C option instead of the `cd` wrappers.

John Public@rss.iscabbs.com — Tue, 21 Feb 2012 16:44:00 -0600

Kena> Use the -C option instead of the `cd` wrappers.
    tar -C from -cf - . | tar -C todir -xpvf -

(Feoh/89904) Just discovered a really nice free GUI client for Git or Mercuri...

Feoh@rss.iscabbs.com — Tue, 21 Feb 2012 10:02:00 -0600

Just discovered a really nice free GUI client for Git or Mercurial on OSX.
Atlassian SourceTree: http://www.atlassian.com/software/sourcetree/overview

(Kena/89903) Oh, entirely. SSH is the way to fly -- but not locally. (My "F...

Kena@rss.iscabbs.com — Tue, 21 Feb 2012 09:23:00 -0600

Oh, entirely.  SSH is the way to fly -- but not locally.  (My "FAQ" file that I
keep updated still refers to using rsh.  Gives you an idea of how long I've
been using this trick.  Though not with modifying the data on-the-fly which,
for the record, does appear to work.  I do wonder a little bit about the tar
metadata, but I imagine a well-named variable would be very unlikely to have a
namspace conflict.)

(Cyanide/89902) I use the tar ssh thing to backup, or take data off servers when...

Cyanide@rss.iscabbs.com — Tue, 21 Feb 2012 08:59:00 -0600

I use the tar ssh thing to backup, or take data off servers when im out of disk
space to make a local backup..

(Egregious/89901) >cd fromdir; tar cf - . | (cd todir; tar xvf -)

Egregious@rss.iscabbs.com — Tue, 21 Feb 2012 08:48:00 -0600

>cd fromdir; tar cf - . | (cd todir; tar xvf -)

You can even put an ssh in there.  Fun way to roll your own rsync.

(Kena/89900) Odd: it is GNU-speicifc. Even relatively recent; perhaps the pa...

Kena@rss.iscabbs.com — Tue, 21 Feb 2012 08:36:00 -0600

Odd: it is GNU-speicifc.  Even relatively recent; perhaps the past seven or
eight years.  The way I used to do it before that was either rsync, or tar.
ILDE MUSING THAT SHOULD BE TESTED BEFORE BELIEVING:
You can use tar to do this via pipes, like this:
cd fromdir; tar cf - . | (cd todir; tar xvf -)
I wonder if you could have a little fun with that, being that you also want to
do replacements of variables for values:
cd fromdir; tar cf - . | perl -e 'while (<>){s/foo/bar/g; s/biz/bang/g;
  print;}' | (cd todir; tar xvf -)

(JL/89898) WINNING!

JL@rss.iscabbs.com — Tue, 21 Feb 2012 04:30:00 -0600

WINNING!

(Danix/89897)

Danix@rss.iscabbs.com — Tue, 21 Feb 2012 04:22:00 -0600

<?php
  echo "up, up, down, down, left, right, left, right, B, A, Select, Start";
?>

(JL/89896) So I've been continually working on this "Game Runner" idea I go...

JL@rss.iscabbs.com — Tue, 21 Feb 2012 04:07:00 -0600


So I've been continually working on this "Game Runner" idea I got a while ago.

At work, we ran our first competition with Tic Tac Toe. The whole thing had
some quirks, and it didn't go well.

I've learned, and cleaned everything up.

NEXT GAME: Connect 4!

Would anyone be interested in writing players and having a little competition
in this forum to see who can write the best player algorithm?

I have example players in PHP, JS, and JSP. Each one tries to make a random
valid move. You would need to:
 1) Modify the code to have some intelligence
 2) Host your file somewhere that it can be run and be called by my controller

It's all documented fairly well.


I have a player that beats the "random" players 99 times out of 100. Good, but
not great.

Anyone interested in trying it out?

(PS - looking up any kind of algorithm in solution ideas on the web is
cheating!)

(Darkhaven/89895) xpost from WWW>

Darkhaven@rss.iscabbs.com — Mon, 20 Feb 2012 10:36:00 -0600

xpost from WWW>
Feb 20, 2012 08:46 from Darkhaven
this is annoying me something fierce. i can't get an event handler to attach to
the list that gets generated with jquery-ui autocomplete. i've tried doing this
both with autocomplete's select method and with a regular jquery .live method,
and (because i was pissed) tried them both together as is shown below. however,
i can't get any handler working here:

    var cwJQ = jQuery.noConflict();

    cwJQ(document).ready(function() {

        $input = cwJQ('input#last_name');

        var renderItemOverride = function (ul, item) {
                return cwJQ("<li></li>")
                        .data("item.autocomplete", item)
                        .append( item.label )
                        .appendTo(ul);
        };

        $input.autocomplete( {
                source: function( req, res ) {
                        cwJQ.ajax( {
                                url: 'url',
                                dataType: 'html',
                                data: ({  "name": $input.val(),
                                        "name_in_id": 1,
                                        "fn_ln": 1,
                                        "dist": 1
                                }),
                                success: function(data) {
                                        var datum = new Array;
                                        cwJQ(data).find('li').each( function(
i, val ) {
                                                datum.push( "<span id='" +
val.id.split('_')[1] + "' name='li-span'>" + cwJQ( val ).text() + "</span>" );
                                        } );
                                        res( datum );
                                }
                        } )
                },
                select: function( event, ui ) {
                        alert();console.log();
                        addMemDat( ui.item.id, ui.item.value );
                },
                minLength: 2
        } ).data('autocomplete')._renderItem = renderItemOverride;

        cwJQ("#li-span").live( click, function() {
                console.log( this );
        } );
    });


ah, just noticed addMemDat() is in there - the handler works just as well
without that function there. it seems to be the .ajax call that is messing this
up and i'm not sure how to write a simple test case with that call (maybe a
google search if the test is needed to get feedback on this). i also had to
override the _renderItem method so that i could pass html (because i have
distinct 'id' but might not have distinct names to reference against).

(Odd/89894) JL, TAP> cp -a does recursive copying with all permissions and...

Odd@rss.iscabbs.com — Sun, 19 Feb 2012 23:53:00 -0600

JL, TAP>   cp -a does recursive copying with all permissions and attributes
kept, as much as possible.  Usually does "what you want".

No clue if that's gnu-specific or not.

(Danix/89893) Faunus> I believe it is mostly because of the whole OO paradigm....

Danix@rss.iscabbs.com — Sun, 19 Feb 2012 17:46:00 -0600

Faunus> I believe it is mostly because of the whole OO paradigm. It gives us
the tools to have type-checking, but also gives some leeway to have
dynamically-created stuff as well. Maybe.

(The After Party/89892) JL

The After Party@rss.iscabbs.com — Sun, 19 Feb 2012 11:58:00 -0600

JL

#!/bin/bash

cp -rp $1 $2
# where $1 is the template dir, $2 is the new dir

find $2 -type f | xargs perl -pi -e 's/%var1%/value1/g'
# search thru $2 new dir, use perl to replace %var1% with value1

(JL/89891) I have a Task>

JL@rss.iscabbs.com — Sun, 19 Feb 2012 11:02:00 -0600


I have a Task>

I want a simple "build script" that should probably be in Perl. My Perl is not
just rusty. It's all but gone.

I want to copy a "template" directory - every file, every directory. For each
file, I want to do a global replace of %var1%, %var2%, etc with specific
values. I want the final result to be a new directory with the
search-and-replaced files, matching the template directory structure.

Is there a tool already out there that is simple that will do this for me?

Perhaps a Perl module that will simplify the task?

I don't want to jump into a full ANT script or whatever. I have no clue how to
use ANT. Maybe I should, but now is not the time to learn.

Any tips? Thanks!

(Faunus/89890) Does it matter that type-checking and certain kinds of object or...

Faunus@rss.iscabbs.com — Sun, 19 Feb 2012 08:40:00 -0600


Does it matter that type-checking and certain kinds of object orientation so
often go together?

An early-bound, heavily type-checked language doesn't *have* to be object
oriented -- see Haskell, which has type classes but they're not a central
feature of the language.  And object oriented languages obviously don't have to
be late-bound and not-type-checked; Smalltalk was specifically designed for
"extreme late binding," in Kay's words, not early and careful type checking.

Why is OO *the* way to get type checking in popular languages these days?
Historical accident?  Or are there some good reasons in there?

(Odd/89889) KalTorak> I always use trict. Doesn't prevent any of the abov...

Odd@rss.iscabbs.com — Sun, 19 Feb 2012 01:21:00 -0600

KalTorak>   I always use trict.  Doesn't prevent any of the above issues (in
perl).  No checking of (number of / type of) arguments, no checking if a string
will ever be initialied before use, etc.

                      ^^^strict^^^

(John Public/89888) I see a correlation with no causality: scripting/dynamically/wea...

John Public@rss.iscabbs.com — Sat, 18 Feb 2012 20:33:00 -0600

I see a correlation with no causality: scripting/dynamically/weakly-typed
languages tend to have more "expressive freedom", and type-safe/
strongly-typed/ compiled languages tend to "feel" more procedural.  The
correlation is between the expressive fun of the language and the type-safety,
yet there's no reason for that to be so... type checking doesn't prevent
expressive syntax. In fact, type-checking is still performed in
dynamically-typed languages; just ad-hoc, many times over and over again, and
way too late to matter.  So there's a disconnect somewhere.

(KalTorak/89887) use strict;

KalTorak@rss.iscabbs.com — Sat, 18 Feb 2012 08:17:00 -0600

use strict;

(Odd/89886) TAP> taking it a bit broader from "type checked" to "compiled i...

Odd@rss.iscabbs.com — Sat, 18 Feb 2012 03:21:00 -0600

TAP>  taking it a bit broader from "type checked" to "compiled in advance":
here's some things I run into with matlab and perl that I've not really
encountered (or differently) in C++:

- empty strings or undefined strings being used further down the program flow,
because they were never filled above.

- truncating floats to int (or worse, 0, in fortran)

- integers that suddenly become floats (or worse, strings)

- not knowing whether your int, read from an xml file, is in cell, string or
numeric form (matlab)

- passing too few parameters and only finding out at runtime

- Finding out about inaccessible code at run-time, or not finding out at all

- not returning a value even though the calling code expects it

- Using variables before they are declared, allocated or initialized

I'm not sure if these kind of problems don't occur to me in C++ because I'm
more experienced in it, or because of the type checking or because of the
compile time checking.  I have a feeling the last one is the most important.
Anyway, I've ran into this LOADS of times, and it means for me operational
programs are MUCH better written in C++, simply because there are no unexpected
run-time errors.  I think using perl more would remove some of the problems,
but not all of them

(Mandel/89885) SFTD, but I guess a more general answer to "What does type check...

Mandel@rss.iscabbs.com — Fri, 17 Feb 2012 18:18:00 -0600

SFTD, but I guess a more general answer to "What does type checking provide?"
would be that it lets you play around on a level one step high than traditional
object oriented languages.  Or maybe even a few levels higher.  Mostly because
yes, all of that code that was written down below can be (should be?)
guaranteed correct.  I guess that's my real answer.

(Mandel/89884) TAP>> I guess one example of static type checking and it's meri...

Mandel@rss.iscabbs.com — Fri, 17 Feb 2012 17:56:00 -0600

TAP>>  I guess one example of static type checking and it's merits (or
demerits) is OCaml's structural sub-typing.  This is my favorite toy language
and one I've still yet to do anything really serious with, but basically
one example would be how you don't have to explicitly define complex types.
A "module" can produce a composite object without a class definition, or
from multiple other classes, or even produce classes dynamically, and the
resultant object either fits, or it doesn't fit into future calls.

let my_func x = x#member_func

Is effectively a function that will take any object that implements
member_func.  It doesn't care what type of object it is.  There's lot of
other funky stuff going on in OCaml like that due to the fact that compiler
is just being smart about types.  Haskell's Type Classes are somewhere in the
same realm as OCaml's Module system, and this will forever be an annoying
debat. :)

(Danix/89883) Heh. I've run into a shitload of code that instead of doing

Danix@rss.iscabbs.com — Fri, 17 Feb 2012 08:35:00 -0600

Heh. I've run into a shitload of code that *instead* of doing

Logger.getLogger("MEH").log(Level.FINE, "This is some debugging message");

actually do

if (debug)
   System.out.println("MEH::MEH This is some debugging message");

.... then they proceed to leave the debug flag on when deploying to PRODUCTION.

Type-checking languages> Validating "old" on an int is pretty much needed in
web environments, or any environment where you're getting untrusted input.
By "untrusted" I mean user-inputted, received by someone calling a WebService,
RPC, sent by mail, whatever. A whole hell of a lot of hacks are made because
some idiot didn't do such a check and the program crashes.

Java, specifically, has much more than just type checking on that matter; JSF
will specifically let you put up format and data validation stuff so that
someone putting age -20 will result in an error stating "dude, you can't be
younger than 0!" or "Silly, February doesn't have 30 days!". That kind of
facilities make me offload the whole data input validation to a predefined,
implemented routine and focus on the business stuff itself. It actually helps
solving the "tight coupling" thingy, though that has more to do with making
your stuff reuseable in other parts of your code. For example, implement a
"Crypto Data Store" which does all the encryption stuff, and presents you with
a "Data Store" where you can store/retrieve stuff. Then use said "Crypto Data
Store" for sensitive user credential data, then for sensitive parameters ,etc
.... you don't have to worry on *where* you're using it, as the implementation
for that Data Store is the same, wherever you use it.

(John Public/89882) yes, it'd be moot. the test is wasteful noise.

John Public@rss.iscabbs.com — Fri, 17 Feb 2012 06:16:00 -0600

yes, it'd be moot. the test is wasteful noise.

(The After Party/89880) JP

The After Party@rss.iscabbs.com — Thu, 16 Feb 2012 18:04:00 -0600

JP

  Hrm, well, considering we ALWAYS have logging (error level for sure,
  and there are always some errors), would the perf gain be moot?
  I really want to know how the in hell this got into the coding standard
  (i assume its there).

(John Public/89879) TAP> You don't have to check the level first. They only do that ...

John Public@rss.iscabbs.com — Thu, 16 Feb 2012 15:39:00 -0600

TAP> You don't have to check the level first. They only do that for
performance reasons, so they can avoid the operation which build the log
message in the case where it will be immediately discarded inside Log4J
anyway.

(Harbinger/89878) TAP>

Harbinger@rss.iscabbs.com — Thu, 16 Feb 2012 15:30:00 -0600

TAP>
You are not crazy.  The entire point of using something like Log4j is not to do
the thing you described...

(The After Party/89877) log4j

The After Party@rss.iscabbs.com — Thu, 16 Feb 2012 14:28:00 -0600

log4j

  Am I crazy?
  logDebug("msg");
  should NOT require a

  if(isLoggingDebug()) { logDebug("msg"); }

  Should it?

(The After Party/89876) Typed Langs

The After Party@rss.iscabbs.com — Thu, 16 Feb 2012 14:23:00 -0600

Typed Langs

  What does type checking provide? It is a layer of validation that your data
  is "more" correct. i put "more" in quotes because while
  int age;
  will prevent "old", it won't prevent -20 or 1000 (if we're talking about
  people that are supposed to be alive).

  (I guess that was JP's point, just giving an example I suppose).

Feoh

  "one off process automation" seems like an oxymoron.
  process automation implies you're gonna be running the thing often.
  So, its not really "one off", at least not in use.
  Maybe you write it once and only use it for one process, so it has
  not reuse for other areas....i can kind of see that. But
  time and time again, and I think i do the same (similar) job as
  you, i find myself wanting some 'base' functionality...maybe not
  the same jobs per say, but i want: logging, email notification,
  progress indicators, log rotation, log cleanup, etc
  I manage java, but I don't write my tools in java (I do wonder if
  that is a good choice from a reuse perspective), but since
  this discussion is about type checking..(or OO?)...not sure
  strong types would make a big diff...OO might.

(John Public/89875) I wasn't limiting my comments to web programming; just observing...

John Public@rss.iscabbs.com — Thu, 16 Feb 2012 11:21:00 -0600

I wasn't limiting my comments to web programming; just observing the
connection between the web and scripting languages.

I'm not arguing for java.  I'm arguing that type checking is an important
cost-saving feature, and the decision to use a programming language without
type checking should be made with that factor in mind.

(Feoh/89874) So were you restricting the scope of your comments just to web p...

Feoh@rss.iscabbs.com — Thu, 16 Feb 2012 11:12:00 -0600

So were you restricting the scope of your comments just to web programming? If
so, then perhaps the light weight kind of process automation work I do really
IS 'scripting' and makes sense in a dynamically typed language, even for you?
:)

I've done programming tasks with lots of external subprocess control in Java.
It isn't pretty :) Ruby handles the kind of one off process automation work I
do *much* more readily.

(John Public/89873) Feoh> Hrmm, I dunno. It's pretty extreme to say there is NO plac...

John Public@rss.iscabbs.com — Thu, 16 Feb 2012 10:39:00 -0600

Feoh> Hrmm, I dunno. It's pretty extreme to say there is NO place at all for
dynamically typed languages. I do enjoy the more quirky languages like lisp
and bash. They're fun, and maybe part of the fun is because they are so risky.
I guess the problem in question emerges when you have dynamically-typed
languages that aren't very different than their strongly-typed counterpart.  I
would opt for the choice that comes with a type-checker.

I do think the myths about dynamically typed languages are finally starting to
unravel.  For example, a lot of programmers say they prefer the lack of type
checking because they feel more productive.  However, there is mounting
evidence that individuals, engineers in particular, are terrible at estimating
their own productivity, and hopeless at estimating the overall team's
productivity.  Our estimates turn out to be based on various emotional
factors, such as how "busy", hurried, or under pressure we feel. Being "busy"
all day, i.e., typing a lot of code, feels productive, but being busy does not
necessarily correlate to overall team productivity.

Another myth is that dynamic typing solves the problem of "tight coupling".
(It sounds idiotic, but I see this argument come up from time to time.)
Coupling has nothing to do with types, so this argument really makes no sense.

It's a shame people have equated so much of the web with "scripting". Maybe
it's all the string-iness of every web protocol and format. The formats and
protocols are all very loosey-goosey, and our brains make the curious mistake
of thinking that the tools we use to work with loosey-goosey data must be
loosey-goosey themselves.

(Feoh/89872) I'm with you Faunus, I've been in exactly the same boat. But, i...

Feoh@rss.iscabbs.com — Thu, 16 Feb 2012 10:59:00 -0600

I'm with you Faunus, I've been in exactly the same boat.  But, if I'm
understanding John Public correctly, he's saying we're misguided, and that ALL
programming should be done in environments with strict time checking,
preferably at 'compile' time :)

(Faunus/89871) Not java. HASKELL, silly. Gotta be hard core about type checki...

Faunus@rss.iscabbs.com — Thu, 16 Feb 2012 09:56:00 -0600

Not java.  HASKELL, silly.  Gotta be hard core about type checking!

Seriously, someday I'm going to have a reason to do actual work in a language
with real type checking.  I think it'd be really cool to learn.  Just never had
a need for that in the (comparatively lightweight) programming jobs I've
actually had.

(Feoh/89870) So are you saying that dynamic languages have no place at all in...

Feoh@rss.iscabbs.com — Thu, 16 Feb 2012 09:27:00 -0600

So are you saying that dynamic languages have no place at all in the world and
that everyone should just run Java and be done? I'm not necessarily saying I
disagree with that, mind, but my personal sphere is so narrow that I'm guessing
even you wouldn't argue that Java is the right tool for my particular job (I'm
a release engineer, and pretty much just write process automation/glue
scripting code, for which Java is a really poor fit :)

(John Public/89869) JL> I don't think anyone is debating whether it can be done. ...

John Public@rss.iscabbs.com — Thu, 16 Feb 2012 09:08:00 -0600

JL> I don't think anyone is debating whether it *can* be done.  If you throw
enough time, money, people, and sweat at it, you'll eventually get something
you can point to and call it "software". The question is not whether you CAN
accomplish something the hard way. It's whether there are better ways, and
whether the right tool or decision will free up resources so you can
accomplish even more.

(Guardian Angel/89868) Facebook isn't just PHP + MySQL. In fact, most of what they're d...

Guardian Angel@rss.iscabbs.com — Thu, 16 Feb 2012 09:10:00 -0600

Facebook isn't just PHP + MySQL. In fact, most of what they're doing on the
backend anymore ISN'T Mysql, it's HBase.
:)

(JL/89867) John Public:

JL@rss.iscabbs.com — Thu, 16 Feb 2012 09:03:00 -0600

John Public:
>Smoke Eater> I agree. But what if the job was initially a small web app, for
>which PHP (or ColdFusion, etc.) may have been an adequate tool, yet over time
>the requirements for the app have grown to the point that template scripts are
>no longer appropriate?

Facebook = PHP + mySql

They've scaled bigger than any other site on the web.

Of course, the customized the hell out of both PHP and mySql, but it's an
example of scaling huge with what you might consider "inadequate" tools.

They aren't a model of development, obviously. But they show that it can be
done.

(John Public/89866) Feoh> It's one thing to vent about annoying syntax quirks betwee...

John Public@rss.iscabbs.com — Thu, 16 Feb 2012 08:36:00 -0600

Feoh> It's one thing to vent about annoying syntax quirks between languages,
and another thing altogether to debate the real effects that decisions have on
the cost or duration of the dev/release cycle, overall product quality,
numerical correctness, flexibility of the software to adapt over time, etc.


We ship some shitty products and some good products. Just because we managed
to ship some shitty products doesn't validate the process or tools used to
build them.

(Feoh/89865) I think there's a place in the world for dynamic programming lan...

Feoh@rss.iscabbs.com — Thu, 16 Feb 2012 08:42:00 -0600

I think there's a place in the world for dynamic programming languages.  There
are situations where you're absolutely right, a compiled language with strict
type checking is the *only* way to fly.  But for a huge swath of problem sets,
that kind of strictness just isn't required.  Sure, programmers have to build
assertions into their code so that when they're passed a foo they check that it
*really is* a foo, but IMO that's a reasonable trade off in ROI terms for lots
of cases.

I *like* Java, but IMO the rise in popularity of languages like Python, PHP,
and Ruby along with their respective web frameworks indicates to me that in the
real world, for a *lot* of cases, getting it done fast and having it produce
*reasonable* results for 90% of cases is more important to people than spending
many extra man months coding boiler plate Java and getting that extra 10%.

(John Public/89864) Feoh> Unit tests have one serious problem in that nobody writes ...

John Public@rss.iscabbs.com — Thu, 16 Feb 2012 08:18:00 -0600

Feoh> Unit tests have one serious problem in that nobody writes them! :)
Joking aside, the stock answer essentially advocates that the programmer spend
time writing unit tests to do what a compiler would already do.  Certainly
even the unit test die-hards would agree that that's not a good way to spend
time.

(Feoh/89863) What if they like working in PHP?

Feoh@rss.iscabbs.com — Thu, 16 Feb 2012 08:23:00 -0600

What if they like working in PHP?

As I've grown older I've come to realize that much of the venting and spewing
about programming language A versus programming language B is, if you'll
forgive me for saying so, a steaming pantload.

After I discovered Ruby I used to rant and roar about Perl, but you know what?
It's a perfectly reasonable tool for a whole swath of tasks.  I went to work at
a PHP shop, and heard a whole bunch of PHP hate.  But you know what? We ship a
rather capable suite of PHP software that runs some of the biggest poltiical
campaigns on planet earth, it works and it works well.

I'm all for 'tool to task' but I've come to believe that picking a programming
language for a project depends on a whole bunch of factors, and is *never* as
simple as people make it out to be.

I don't love PHP, I'm hardly a fan boy, and there are times when I work with it
that I think to myself Are you serious? REALLY? But that has exactly zero
impact on its usefulness for the task at hand.

(Smoke Eater/89862) Retraining on whatever the tool is. Invest in your people and t...

Smoke Eater@rss.iscabbs.com — Thu, 16 Feb 2012 08:21:00 -0600

Retraining on whatever the tool is.  Invest in your people and they'll invest
in you.

(Feoh/89861) Or, if you have a shop full of PHP programmers? I'm not convinc...

Feoh@rss.iscabbs.com — Thu, 16 Feb 2012 08:12:00 -0600

Or, if you have a shop full of PHP programmers?  I'm not convinced that PHP is
a horrible choice for OO web applications actually.  Again, yes it's ugly, but
that doesn't make it a poor choice.

John Public> WRT dynamic languages - the stock answer is that that's what unit
tests are for.

(John Public/89860) Smoke Eater> I agree. But what if the job was initially a small ...

John Public@rss.iscabbs.com — Thu, 16 Feb 2012 07:56:00 -0600

Smoke Eater> I agree. But what if the job was initially a small web app, for
which PHP (or ColdFusion, etc.) may have been an adequate tool, yet over time
the requirements for the app have grown to the point that template scripts are
no longer appropriate?

(Smoke Eater/89859) Just because you can do something, doesn't mean you should. I...

Smoke Eater@rss.iscabbs.com — Thu, 16 Feb 2012 07:45:00 -0600

Just because you *can* do something, doesn't mean you should.  I'm a big
believer in fitting the tool to the job, especially when choosing a different
tool can be free.

(John Public/89858) Feoh> A problem with a large system in PHP (or other dynamically...

John Public@rss.iscabbs.com — Thu, 16 Feb 2012 07:16:00 -0600

Feoh> A problem with a large system in PHP (or other dynamically typed
languages) is that it becomes difficult to maintain or refactor, because you
can't get any compile-time checks to prevent regressions. Faunus won't know if
he breaks something until very late in the cycle. A little typo could be
potentially very costly in terms of the number of person hours spent before
the thing is detected.  I see it as make-believe OO: The language syntax
support for OO is all there, but none of the assurances are.


Faunus> You know I love performance optimization or rearchitecture. But
there's always a human standing in the way of it. There's always someone ready
to whip out the "80-20" rule, and you bet they'll be sure to define whatever
it is THEY want to be in the "80%" zone, and whatever it is YOU want to be in
the 20% zone.  That's a battle you can rarely win, because the terms are
defined in their favor from the start.

I've found that nobody appreciates performance optimizations (or
rearchitecture) unless the optimization is,

    (a) big enough that it opens up new possibilities, and
    (b) already done! :)

The first means you have to think outside the "80-20" crap.  The second means
you should never propose a performance enhancement. Just do it and take credit
afterwards!!

(Faunus/89857) No doubt there. I'm just concerned whether the classes I'm deal...

Faunus@rss.iscabbs.com — Thu, 16 Feb 2012 06:18:00 -0600


No doubt there.  I'm just concerned whether the classes I'm dealing with make
any sense. :)

(Feoh/89856) Our whole suite is object oriented PHP. Sure, it's ugly, but an...

Feoh@rss.iscabbs.com — Thu, 16 Feb 2012 05:58:00 -0600

Our whole suite is object oriented PHP.  Sure, it's ugly, but anyone who thinks
you can't create sizable OO systems in PHP hasn't tried.

(John Public/89855) Faunus> Sounds like one of those cases where you've connected A ...

John Public@rss.iscabbs.com — Wed, 15 Feb 2012 20:00:00 -0600

Faunus> Sounds like one of those cases where you've connected A to Z, and
you've realized you can strip out everything in between. Don't hesitate, just
cut out all those dumb middle layers. They may have crept into the design over
time, arguably justified at each step, but that doesn't matter one iota
anymore.

(Faunus/89854) Yeah. It'd have to be several prepared statements, cause the ob...

Faunus@rss.iscabbs.com — Wed, 15 Feb 2012 17:11:00 -0600


Yeah.  It'd have to be several prepared statements, cause the objects
are complex and aspects of them appear in several tables in the local
database.

I'm glad it's not just me.

Oh, it's PHP, BTW. :)

(The After Party/89853) Faunus

The After Party@rss.iscabbs.com — Wed, 15 Feb 2012 15:59:00 -0600

Faunus

  I agree, its kinda weird.
  I agree with your approach.
  One class that connects to the old data source, creates an array/list of
  the objects that represent the old data, connect ot the new db, write
  them (persist) by looping thru your array/list, probalby using a prepared
  statement..

(Faunus/89852) OK I wrote and rewrote a couple bits of that and said things tha...

Faunus@rss.iscabbs.com — Wed, 15 Feb 2012 14:33:00 -0600


OK I wrote and rewrote a couple bits of that and said things that don't quite
make sense like saying "call the instantiated Importer i" and then never
referring to "i" again but hopefully it's comprehensible.

(Faunus/89851) Oy. I'm in one of those situations where I don't know if I'm se...

Faunus@rss.iscabbs.com — Wed, 15 Feb 2012 14:00:00 -0600


Oy.  I'm in one of those situations where I don't know if I'm seeing
something kinda stupid or if I just don't know enough for it to make
sense to me.

I was looking at some object-oriented code today in a language I shall
not name so as not to prejudice anyone.  Some of this seems kind of
stupid to me.

The idea here is we go through a list of items from a remote database
using a http-based (SOAP) remote API.  First we grab the list, and
then we go through the list one by one and grab some chunks of info
about that item, then we reformat those chunks of info and put them
into our new database.

We're limited to accessing the remote DB via this shitty API, because
of reasons.

Anyway, this is the "OO" way we do it.

Here's the first class: Importer.  Because of local architecture
limitations, this has to be a class from which, when we press a
button, one object is instantiated and one method is called on that
object.

What the Importer object (call it i) does first is grabs the long list
of items in the remote DB via HTTP.

Then it iterates through that list, and for each item, it goes through
and grabs a bunch of chunks of information about that item and stores
them *in its own fields* (fields, properties, whatever you call them.
Its own local variables accessible via get/set methods.)

So at any given moment, it has a bunch of properties which are *really*
the properties of whatever item it is currently gathering information
about from the remote database.

That is just conceptually *weird* to me.  Making the same object be
the one doing the iterating, and also in some way assume the properties
of the things it's iterating over?  Makes me go eww.  (Also seems
dangerous.  What if a bug keeps some property from being reset from
one item to the next, so that properties are "left over" from the
previous item in the list?)

Second class is the Formatter.

What the Importer does after it is satisfied that it has re-created
itself in the image of the currently iterated item, is that it creates
a new Formatter and it hands itself to the Formatter in the
Formatter's constructor.

The Formatter's constructor pulls information out of the Importer, and
munges it up into a new data structure that it stores inside itself.  All
of this happens inside the constructor.

We store a reference to the newly-created Formatter in our Importer
(overwriting the last Formatter we created, which gets garbage
collected) so we can hand it over to the constructor of our third class.

The Third class is the Inserter.

The Inserter's constructor is handed the Formatter and a reference to
a local DB.  In the constructor, the Inserter retrieves the data
object that the Formatter had inside itself, and inserts rows into
several local database columns.  Though the Inserter is never used
again, it is nevertheless stored in the Importer until it is
overwritten the next time we create an Inserter.

Now we return and iterate again, creating a new Formatter for every
item in the remote database, and handing it to a new Inserter, and
storing those untilt hey're overwritten on the next iteration.

Does that make sense at all?

I don't even know if I'd use objects for most of this.  (The language
in question has optional object orientation, not mandatory like Java.)

If I did, I'd probably have an class of object which represents one of
those items in the list.  They're products, call it Product.  It
wouldn't be any fancy ActiveRecord shit, it would be just for the
purpose of this import.  Call it an ImportedProduct to make it clear,
say.

I'd have the Importer grab the list of products in the remote DB
and create and ImportedProduct from each.  The ImportedProduct would
be smart enough to munge its own data into a format suitable for the
local DB.  I'd probably let the Importer do the inserting too, since
it's got a reference to the database anyway and it's doing the
iterating.

So two classes, one which we are required to have by our architecture,
and one which is useful because it encapsulates a chunk of data and
allows access to it in a different format than it went in.

You could do a lot of little variations on this, like making those
objects smart enough to remote-query themselves into existence, or
smart enough to write themselves into the local DB if you hand them a
reference to it.  Whatever.

But having the Importer change its own properties while it iterates?

Having a class that doesn't do anything but run its constructor?

Having another class you create and then hand it a copy of yourself so
it can query you for various values you happen to currently hold as
you iterate through a list of other things?

It's not totally insane (I thought it was at first) but still it just
seems kind of *off*.

Is this one of those "Design Patterns" I keep hearing about?

(John Public/89850) That is unfortunate, but java doesn't do that for static final...

John Public@rss.iscabbs.com — Tue, 14 Feb 2012 14:25:00 -0600

That is unfortunate, but java doesn't do that for *static* final data. Is
there a reason you don't want to make it static?

There are a couple other terrible problems with java's class file format. They
apparently tacked on `long` and `double`, and yet 32-bit assumptions are all
over the format.  "references" (i.e., pointers) are expected to take 32-bits
in the _runtime_ operand stack, or half the units of `double` or `long`.  The
op stack units are not 64-bit aligned, yet contain 64-bit values (such as
double and long).  There are several untyped op codes that require the runtime
to magically know the types of the values on the stack (Have a look at
`dup2_x2` in the spec http://tinyurl.com/7a8efcq )

And if you ever get in one of those debates about whether Java is really
"compiled" take a look at what the runtime loader must do for every method
(http://tinyurl.com/874qhty) Did you catch it? Full data flow analysis AND
symbolic interpretation. Serious compilers perform one or the other, but not
both, and scripty languages don't perform either. So is java "compiled"?  You
bet your ass it is! Every single time it's loaded, that is. Java's slogan
ought to be,

    Compile Once, Compile Everywhere!

(Ahto/89849) One feature of the Java spec that I find rather annoying is that...

Ahto@rss.iscabbs.com — Tue, 14 Feb 2012 10:57:00 -0600

One feature of the Java spec that I find rather annoying is that the values of
public final fields are copied into client classes when the client classes are
compiled. The result is that chaning the value and re-compiling the containing
class will leave different parts of code with different values for the field.

(Lochner/89848) I feel the same way about C vs. Java. I have a very intimate un...

Lochner@rss.iscabbs.com — Tue, 14 Feb 2012 02:41:00 -0600


I feel the same way about C vs. Java.  I have a very intimate understanding of
what all is going into my C program, and when something blows up, I know how to
track it down and debug it, I know where to look.

These enormous, complex Java compilers and IDEs?  Shit, I don't even know where
to begin.  That might be lack of familiarity rather than the IDE, but there is
someting about having that fine-tuned control of C that really appeals to me.

I've done pointers for so long that I don't know how to code without them.

(Danix/89847) Heh. My recent re-introduction to C programming made me realize ...

Danix@rss.iscabbs.com — Tue, 14 Feb 2012 02:12:00 -0600

Heh. My recent re-introduction to C programming made me realize how much I had
grown used to the Java model. Most of Java's intricate stuff is hidden by
modern IDEs, while C programming I still do using vim and Makefiles. Thus I'm
actually savvier on what stuff my C module is linking against, while most of my
Java stuff is more of a "ah, add this JAR to the classpath so the errors go
away" thing.

I've also done C#, which basically does the Java model language-wise, but uses
the DLL/shared-library stuff at link time. Oh, both Java and C# *hate* when you
get overlapping JARs/DLLs and won't usually tell you what happened! Programs
will barf at runtime, and you have to manually check the NoClassDef exceptions
to check if it is a case of duped classes. :(

(Cyanide/89846) It isn't necessarily a step away from the machine any longer..

Cyanide@rss.iscabbs.com — Mon, 13 Feb 2012 14:17:00 -0600

It isn't necessarily a step away from the machine any longer..
http://www.arm.com/products/processors/technologies/jazelle.php
Jazelle technology is a combined hardware and software solution from ARM. ARM
Jazelle technology software is a full featured multi-tasking Java Virtual
Machine (JVM), highly optimized to take advantage of Jazelle technology
architecture extensions available in many ARM processor cores.

Sorta like an mpeg/jpeg or encryption engine embedded in hardware..

(John Public/89845) Feoh> I'm cynical as hell, so I think the next step will be to d...

John Public@rss.iscabbs.com — Mon, 13 Feb 2012 12:49:00 -0600

Feoh> I'm cynical as hell, so I think the next step will be to do away with
deployed executable formats altogether. Everything will be just source code
scripts interpreted at runtime. Probably on some future iteration of ECMA
script, which will be indistinguishable from Java 1.0.

I don't think it's a step *forward* by any means, though. The classpath and
scripting language approach is a step AWAY from the machine. And we know what
always happens when something takes a step away from the machine: it creates a
performance gap.  Then, instead of the obvious, sensible, correct solution
(step back towards the machine, duh), people attempt to fill that gap with
more layers.

So my prediction is that everything will gravitate towards the C model, but in
a very roundabout way. It'll be "behind the scenes", buried beneath multiple
virtual layers. It's already happening with .NET assemblies, JIT, re-compilers
like Dalvic, and so on.

(Feoh/89843) Actually I don't think Java's 'link' behavior is particularly tr...

Feoh@rss.iscabbs.com — Mon, 13 Feb 2012 10:07:00 -0600

Actually I don't think Java's 'link' behavior is particularly transparent, for
exactly the reasons you specify.   In that regard, using the CLASSPATH, it's
actually pretty similar to Ruby ($LOAD_PATH), Perl (@incpath I think?) and PHP
(I forget what it calls it :)

An interesting amalgam of approaches happens with Microsoft's C# - it actually
compiles into byte code, which is then packaged up into actual executables like
traditional compilers would do, and libraries are contained in (for that
platform) bog standard shared library packages.

Which IMO is a bad trade-off because then you're stuck in DLL hell as an end
user if your installer does something you're not expecting.

(John Public/89842) Linking model: What I was getting at was how it affects the whol...

John Public@rss.iscabbs.com — Mon, 13 Feb 2012 09:46:00 -0600

Linking model: What I was getting at was how it affects the whole development
process. For example, Java's linking model has the effect of requiring one
file for each (non-private) class, and the filesystem layout is required to
match the package namespace layout.  Every method of every class is a
potential entry point, so there's no clean exported API, and you have no way
to analyze API dependencies. Every member of every java class is runtime
linkable, so this requires the runtime loader to be a monster, performing a
lot of work that could have been performed at compile-time.  To mitigate the
problem of so many files and cross-dependencies the java compiler
auto-recompiles files as it needs to, based on timestamps, so there is no such
thing as compiling a single java file, since it could trigger hundreds more
compilations.  Java's inefficiencies have created a huge market for "JIT"
tools, research, out of thin air.

These are all effects of the link model to some degree.  Do you think these
sort of effects are burdens on the programmer? I'm not trying to single out
Java. I'm  using it as an example of how the link model, however "transparent"
it may be, has many ramifications, affecting the development process and even
the language itself.

(Feoh/89840) So, has anyone ever used a language where closures were used per...

Feoh@rss.iscabbs.com — Mon, 13 Feb 2012 08:56:00 -0600

So, has anyone ever used a language where closures were used pervasively
throughout the standard libraries?

I ask because in Ruby, blocks are used *everywhere* and some folks here are
horribly down on the language, so I was wondering if there were others I should
look into.

(Feoh/89839) John Public> This may sound like I'm being facetious but really ...

Feoh@rss.iscabbs.com — Mon, 13 Feb 2012 04:54:00 -0600

John Public> This may sound like I'm being facetious but really I'm not.  The
compile model I like best is the one that's the most transparent to me as a
programmer.  I prefer languages that optimize for propgrammer productivity like
Ruby and Perl and I'd guess Python and others.

I'd probably love LISP for that same reason if I ever took the time to really
absorb it :)

(John Public/89838) Different languages support varying compile-link-load models. c...

John Public@rss.iscabbs.com — Mon, 13 Feb 2012 04:23:00 -0600

Different languages support varying compile-link-load models.  classic C
compiles to object files, and those can be blobbed together into a library,
but they ultimately need to be linked.  Java's model is to compile each unit
directly to the final executable format (the class file), in which every
symbol is exported, and must be linked. The Java runtime must perform all the
linking and loading. Perl does something different (compiles and links the
whole world at runtime). Just to name a few.

What model do you prefer, and why?

(Smoke Eater/89837 Forum Moderator) How to pronounce a word is best discussed in Linguistics>, which...

Smoke Eater@rss.iscabbs.com — Sun, 12 Feb 2012 14:21:00 -0600

How to pronounce a word is best discussed in Linguistics>, which is down the
hall...

(River Rat/89836) How do you guys pronounce data?

River Rat@rss.iscabbs.com — Sun, 12 Feb 2012 14:19:00 -0600

How do you guys pronounce data?

dey-tuh

dat-uh

dah-tuh

(Stilgar/89835) Different Dev branches per feature team. Only releasable code g...

Stilgar@rss.iscabbs.com — Fri, 10 Feb 2012 08:23:00 -0600

Different Dev branches per feature team.  Only releasable code goes into the
Main branch. When main is published it is branched.  Dev teams are responsible
for keeping their branch in sync with what's in Main.  Each Dev branch has it's
own build for QA.  Main also has builds, but that's for more of a accptance
testing.

Code shared between product teams is branched and built into libraries, so the
shared code might be farther along than the version any given product team is
using in their Main branch.

All of the builds are continuous (and most are gated) so code that blows up our
code analysis, unit tests or automated tests never makes it into any branch.

(Feoh/89834) IMO it all comes down to having a really good SCH workflow defin...

Feoh@rss.iscabbs.com — Fri, 10 Feb 2012 06:30:00 -0600

IMO it all comes down to having a really good SCH workflow defined.  We have
that exact situation John Public and the model I described previously works
great:

* One repository per product
* One branch per release
* All contributions start out in 'bug branches' whether it's an actual bug fix
or an enhancement, this bug is identical to the one in JIRA our bug tracking,
and we have hooks in Git to automatically update JIRA when fixes are merged,
etc.
* When bugs are ready they're merged into the release branch by the developer
* We tag RCs on the branch as bugs are fixed until QA gives the nod, then the
branch / release gets its final tag, the release gets shipped/deployed, and
that final tag gets merged back to master (where bug branches are created from)

(John Public/89833) I'm curious how y'all integrate your revision control systems in...

John Public@rss.iscabbs.com — Fri, 10 Feb 2012 06:24:00 -0600

I'm curious how y'all integrate your revision control systems into your
processes. For example, multiple dev teams, working on multiple overlapping
projects (shared libs, etc.), for multiple clients or targets, with various
release schedules, a steady stream of new defects, new features, upgrades,
regressions from QA, last-minute changes to requirements, and prototypes,
etc., you have a process in place that copes with all those factors. How does
your revision control fit into that?

(Stilgar/89832) I've worked on 3 successfully agile teams. It's a happy place t...

Stilgar@rss.iscabbs.com — Fri, 10 Feb 2012 06:15:00 -0600

I've worked on 3 successfully agile teams.  It's a happy place to be.  But yes,
it's counter-intuitive at first.  I think one of the big mistakes I see is that
people bitch about their problems, but they don't do anything to fix them.  You
might find that your build process sucks for agile development...you need to
take the time to adapt rather than jsut throw up your hands.

(Feoh/89831) I think the problem is that once it become a formalized methodol...

Feoh@rss.iscabbs.com — Fri, 10 Feb 2012 05:24:00 -0600

I think the problem is that once it become a formalized methodology, it lost
its essential benefit, or much of it.  If you look at the people who started
the movement, e.g. the Pragmatic Programmers, they just changed they way they
work in order to embrace a few really simple core principles, and that was it.
When I hear people talk about "Agile with High Ceremony" it just makes me
laugh.

(John Public/89830) That's the main problem I hear from Agile teams: Nobody actually...

John Public@rss.iscabbs.com — Fri, 10 Feb 2012 04:38:00 -0600

That's the main problem I hear from Agile teams: Nobody actually does it.  You
need the whole team committed to Agile, or else it all falls apart. The
pattern I see is a couple of dedicated Agile guys who embrace Agile in
earnest, but they are surrounded by a sea of dont-rock-the-boat people who
have made careers out of appearing busy. Even if they somehow get management
sign-off, management remains aloof from Agile, and at the slightest hiccup,
the fingers all point to Agile, and mgmt will override the process and derail
the whole thing. I've seen this happen, geez, numerous times. With small
teams, and with whole shops. Of course I only hear the tale told from the
Agile advocates. But since I know them and in many cases have worked with
them, I trust that I'm getting a relatively accurate picture.

In my offhand experience, Agile's track record is like 0 out of 10. Whether
that's the fault of the pure methodology, or the fault of the team who won't
commit entirely to it, who's to say. Even if nobody seems to have success with
it, I do think the *influence* of Agile is contagious, in a positive way, and
I see specks of Agile here and there integrated into more structured teams.

(Stilgar/89829) I agree with that. When we started agile we were doing QA work ...

Stilgar@rss.iscabbs.com — Thu, 09 Feb 2012 08:39:00 -0600

I agree with that.  When we started agile we were doing QA work along with dev
work and even simultaneously.  The legacy teams told us it was impossbile and
wouldn't work.  They like to say they're aglie, but they really still do
waterfall, miss their ship dates and find critical bugs just before or after
RTM.  Just because you hvae a meeting every day doesn't mean you're doing
scrum.

(Darphbobo/89828) QA> I have seen the good the bad and the ugly when it comes to Q...

Darphbobo@rss.iscabbs.com — Thu, 09 Feb 2012 08:36:00 -0600

QA> I have seen the good the bad and the ugly when it comes to QA.

I am a Senior QA for a large company and I know that the red tape can kill. But
in the long run if your QA team works with dev that is a big thing it can make
or break a team.

Our agile teams work directly with dev every day and it is awesome to see. I do
a mix of Waterfall and Agile methodologies. I prefer Agile becuase our team
communicates so well there is not a lot of guess work.

I have seen the bad to much tape WAY over budget, lots of finger pointing.

the Ugly really is ugly when it happens when code goes out and it is "tested"
and breaks up and downstream.

To me bottom line is that The Devs are awesome they do great work and we check
to make sure that work is the top quality we need it to be. We all have bad
days or off days while coding and that is what QA is here to help check.

(Kena/89827) If you Google my name, the sixth hit is me blasting him for bein...

Kena@rss.iscabbs.com — Thu, 09 Feb 2012 07:45:00 -0600

If you Google my name, the sixth hit is me blasting him for being an arrogant
homophobe-in-hiding (based on a just crazy insane blog post he made).  I mean,
I think he's done some good -- but that's *despite* who he is, not because of
it.  Oh, and, yes, let's not forget the epitome of his writing, which could be
paraprhased as "sex instructions for hackers."

Gosh, what a loser.

(Feoh/89826) Couldn't agree more.

Feoh@rss.iscabbs.com — Wed, 08 Feb 2012 08:33:00 -0600

Couldn't agree more.

I really don't like his writing style either.  And his books on UNIX suck too
:)

(Faunus/89825) esr is like the epitome of every douchey, worthless thing about ...

Faunus@rss.iscabbs.com — Wed, 08 Feb 2012 08:27:00 -0600


esr is like the epitome of every douchey, worthless thing about "hackers,"
everything that would make you ashamed to be one.

(Feoh/89824) Kena> That's why I first mentioned NHD first edition! ESR is....

Feoh@rss.iscabbs.com — Wed, 08 Feb 2012 05:15:00 -0600

Kena> That's why I first mentioned NHD *first* edition!  ESR is.. quite a
character.  Most of my crowd avoids him like the plague at conventions.  The
last time I saw him, he came drunkenly stumbling into a social get together
only party, saying "WHERE'S THE BOOZE AT?"  We aimed him out the door and down
the hall :)

(Danix/89823) Heh, I remember doing programs like that of Mel's story back w...

Danix@rss.iscabbs.com — Wed, 08 Feb 2012 04:54:00 -0600

Heh, I remember *doing* programs like that of Mel's story back when I was
monkeying around with x86 assembly. Self-modifying code! Oh yes!

There was one system, can't remember which where the bootstrap loader would
start loading at [end+x], then loading the boot stuff backwards. The
instruction at [end] would be a "jump to loop beginning", and when the last
instruction was written, it would be written at [end] so the loaded boot sector
would load immediately. Clever stuff...

(Kena/89822) Catching up a little: I enjoy the story of Mel the Hacker, but I...

Kena@rss.iscabbs.com — Wed, 08 Feb 2012 03:34:00 -0600

Catching up a little: I enjoy the story of Mel the Hacker, but I admit that The
Hacker's Dictionary has gone w-a-y down in my estimation since ESR took it
over; his paternalistic, "I speak for the hackers" tone pisses me the fuck off.

Testing: Some things -- even critical things -- can be hard to test.  For
example, at Segway, we didn't have a "virtual Segway" to test all our
subsystems on -- the only way we had to test a full suite was to test ride the
systems, themselves.  Major releases meant an aggregate of 1,000 hours (I think
it was?) of riding before we'd let it out the door, and any aberrant behavior
had to be explained and fixed before it went anywhere.  12.5 MPH doesn't sound
fast until you fall from it.  (Fortunately, each unit has "black box"
functionality of 30 seconds leading up to a critical event, which really helped
on the "explain" front.)  Even so, they still had two recalls, where firmware
had to be updated.

(Smoke Eater/89821) Sorry to interrupt the thread guys, I'm really liking where this...

Smoke Eater@rss.iscabbs.com — Tue, 07 Feb 2012 10:17:00 -0600

Sorry to interrupt the thread guys, I'm really liking where this is going.

I've got a small problem, and it's driving me crazy.

I'm using Miccrosoft Access (2010) to generate a Microsoft Excel (2010)
spreadsheet.

Under VBA, I'm creating the spreadsheet and putting an autofilter on each
sheet's first row (where the column names are).

When I exit Access and open Excel, I'm on the first sheet, and the filter is
there, but frozen...does not kick off unless I tab to one of the subsequent
sheets, then tab back.  Granted, it's a small thing, but it's kind of
important.  Anyone have any ideas?

(Feoh/89820) In some sense, I really admire the community that's grown up aro...

Feoh@rss.iscabbs.com — Tue, 07 Feb 2012 08:49:00 -0600

In some sense, I really admire the community that's grown up around the JVM.
Languages like Scala, JRuby and Groovy, all being able to share code and
libraries is something really special IMO.

Got me to thinking, can you imagine if the entirety of the Ruby, Python, PHP
and Perl communities could release a library that would be out of box usable in
any of those environments?

Anyway, yes, I know I'm daydreaming here.

(Feoh/89819) It appears today is agree with John Public day for me :) Yeah, I...

Feoh@rss.iscabbs.com — Tue, 07 Feb 2012 07:28:00 -0600

It appears today is agree with John Public day for me :) Yeah, I think
quibbling over the semantics of the word is silly.  It's like any job, really.
My wife's a banker, but what she does day to day is manage giant ass IT
projects.  Sure, she works at a bank, but what she DOES is project management.

Another example is "Release Engineer" - that job (which is what I do) can
entail anything from being a mindless lever puller who minds the builds and
whines at people when they implode all the way up to a full fledged product
manager, software developer and infrastructure architect.

(JL/89818) "Programmer"> Eh, it's semantics. I rarely use the word "program...

JL@rss.iscabbs.com — Tue, 07 Feb 2012 06:22:00 -0600


"Programmer"> Eh, it's semantics. I rarely use the word "programmer" anyway.
After all, people "program" their DVR, right?

To me, "programming" just means someone who feeds instructions to a device, or
slightly more advanced, creates algorithms to solve problems.

The word "Programmer" is similar to "Doctor". A nice word to have, but not
specific enough if you have a problem.

(Danix/89817) JP/gSOAP> It seems that the latest stable (2.8) is thread-safe, ...

Danix@rss.iscabbs.com — Tue, 07 Feb 2012 06:07:00 -0600

JP/gSOAP> It seems that the latest stable (2.8) is thread-safe, and at least
I've been able to make it work w/o crashing. I've checked my own code for
memory leaks, haven't checked gSOAP itself to see if its actually freeing its
memory. I had to re-arrange some of the function calls because of the way
they're being called by the main program, so it doesn't really look like the
samples given by gSOAP's docs. While documentation did mention what I had to do
(which fixed the segfaults), it didn't say this was needed in a clear manner; I
had to explicitly search the whole dog for "soap_ssl_init" and
"soap_ssl_client_context" and finally found the tidbit about soap_copy() and
soap_ssl_init having to be called only once.

QA cycles> Productivity booster when done well. Most stuff would go on to QA
and we could continue developing other fixes in the meantime. fixes needed on
the tested code would be fixed on the next release and on the branch we sent to
QA; otherwise the release would go straight into production.

That has been mostly my experience with QA.

(John Public/89816) I don't see the harm in using the same word "programmer" to desc...

John Public@rss.iscabbs.com — Tue, 07 Feb 2012 05:45:00 -0600

I don't see the harm in using the same word "programmer" to describe an HTML
programmer and an assembly programmer. It's no different than "manager". Just
as we recognize many different fields, types, and levels of management, we
recognize that "programming" isn't a single thing.  Or maybe he's just annoyed
that HR and payroll tend to see all programmers alike...  then I'm with him
100%! :)

(Smoke Eater/89815) http://blackhole12.blogspot.com/2012/02/programmers-are-overgene...

Smoke Eater@rss.iscabbs.com — Tue, 07 Feb 2012 05:33:00 -0600

http://blackhole12.blogspot.com/2012/02/programmers-are-overgeneralized.html

Sorry I can't paste the entire text of that article here, but I found it to be
a very interested and thought-provoking article on the state of software
development.  Anyone care to comment?

(Cyanide/89814) Good QA> even if you don't have a whole test suite, and just hav...

Cyanide@rss.iscabbs.com — Tue, 07 Feb 2012 05:04:00 -0600

Good QA> even if you don't have a whole test suite, and just have certain tests
you run to check for known issues and potential problems before deployment
can be a godsend. You can usually think of about 30 of them, then it is
the 31st one that bites you.. :) But you have at least eliminated 30 of the
issues from the list..so it does save time.. :)

(Cragmor/89813) The logic is not working. Or, more to the point, I really am not...

Cragmor@rss.iscabbs.com — Tue, 07 Feb 2012 04:56:00 -0600

The logic is not working. Or, more to the point, I really am not sure what I am
doing when it comes to this part of checking for multiple instances of a
process.

(Feoh/89812) So the logic in your script that detects a dupe isn't working? O...

Feoh@rss.iscabbs.com — Tue, 07 Feb 2012 04:55:00 -0600

So the logic in your script that detects a dupe isn't working? Or processes
appear only once in the process list even if multiple instances are running?

(Cragmor/89811) Yeah, I can list the processes fine. What I cannot do is determi...

Cragmor@rss.iscabbs.com — Tue, 07 Feb 2012 04:48:00 -0600

Yeah, I can list the processes fine. What I cannot do is determine if a process
is running more than once, and if so, killing all instances.

(Feoh/89810) Cragmor> Are you using WMI to get the list of processes? There'...

Feoh@rss.iscabbs.com — Tue, 07 Feb 2012 04:37:00 -0600

Cragmor>  Are you using WMI to get the list of processes? There's an example
for that here:
http://www.computerperformance.co.uk/vbscript/wmi_process.htm

(Cragmor/89809) Need some help. I need a script to run on XP to check for a proc...

Cragmor@rss.iscabbs.com — Tue, 07 Feb 2012 04:33:00 -0600

Need some help. I need a script to run on XP to check for a process to see if
it is running more than once. If so, kill them all, and relaunch once.
I have a script I wrote that will launch if not running, and can use that for
the last part, but I cannot seem to get the first part to work. VBS is what I
am using.

(Feoh/89808) I'm with John Public on this one. REALLY good QA can be a thing...

Feoh@rss.iscabbs.com — Tue, 07 Feb 2012 04:30:00 -0600

I'm with John Public on this one.  REALLY good QA can be a thing of beauty.  I
see it here every day.

(John Public/89807) JL> You are probably skeptical of QA because you've never seen i...

John Public@rss.iscabbs.com — Tue, 07 Feb 2012 03:52:00 -0600

JL> You are probably skeptical of QA because you've never seen it work in
practice.  I've seen QA work wonderfully, and when it does, it multiplies
productivity, boosts pride in the releases, and clients are impressed.  I've
also seen the Massive QA Bureaucracy at the center of the development universe
suck time into its infinite hole.

You're in an enviable position because your team is not already entrenched in
bad QA practices. You don't have to dig your team out of a QA black hole or
fight the overpaid tenants of the QA bureaucracy.  You can start from scratch
and set in motion an efficient and productive testing system.

As for cost analysis, be sure to factor in the cost of defects. How much time
is spent fixing defects? Do production defects incur other costs as well, such
as direct monetary loss, soured relationships with partners, or dissatisfied
customers?  How much of those costs could you "buy back" with automated tests?

(Stilgar/89806) You also don't need to start with full regression testing. You ...

Stilgar@rss.iscabbs.com — Tue, 07 Feb 2012 03:05:00 -0600

You also don't need to start with full regression testing.  You add new tests
as you touch on parts of your solution.  Eventually you will have good
coverage.  Doing TDD isn't really any slower than "wild west" if you're doing
it right.

(Smoke Eater/89805) The problem is that management doesn't give a @#$%@#% about it b...

Smoke Eater@rss.iscabbs.com — Tue, 07 Feb 2012 01:14:00 -0600

The problem is that management doesn't give a @#$%@#% about it being somewhere
in the middle.

(JL/89804) Nevertheless, any step in the direction of setting up more testi...

JL@rss.iscabbs.com — Mon, 06 Feb 2012 17:51:00 -0600

Nevertheless, any step in the direction of setting up more testing is indeed a
step away from what could have been developed with that time. You can't just
magically add in automated regression testing on a 900-screen app without any
up-front cost and commitment.

You can't have it both ways, but you can have it somewhere in the middle.

(John Public/89803) I may be reading into this, but it seems there is an underlying ...

John Public@rss.iscabbs.com — Mon, 06 Feb 2012 13:38:00 -0600

I may be reading into this, but it seems there is an underlying assumption
that testing prevents rapid releases, hinders smart developers, and increases
development time, so eliminating testing will increase productivity.

That assumption is akin to admitting you only know how to test in such a way
that will make things worse. The logic is circular: testing is a costly waste
of time, therefore testing is bad.

I say the team can make testing anything they want it to be. There's no rule
that says you must create a massive QA bureaucracy, as so many shops do.  Or
that you must drink the kool-aid from the latest TDD craze, or commit
whole-hog to refactoring all your code to support a vendor's unit-test
framework.

(JL/89802) I think the biggest threat is not just the fact that production ...

JL@rss.iscabbs.com — Mon, 06 Feb 2012 12:01:00 -0600

I think the biggest threat is not just the fact that production may have a
problem for a couple days, but that DEVELOPERS will get sick of it.

Professional developers have standards, and want to do things in a good way.
Telling them to throw out their proven methodology, and instead subscribe to a
"Wild West" way of developing may really upset them after a while, and
developers may jump ship.

Or not. Who knows. Facebook's insane environment is even more "Wild West" then
a "no rigorous testing" environment, and look where they are.

Clearly, good things can happen when smart people move fast.

I just suspect that most developers are much more conservative than that. I'm
kind of comfortable with the idea. I know others aren't.

(Stilgar/89801) You can have it. I've been there and I'm not going back.

Stilgar@rss.iscabbs.com — Mon, 06 Feb 2012 10:13:00 -0600

You can have it.  I've been there and I'm not going back.

(Smoke Eater/89800) Which is exactly why he feels safe with the attitude of 'fuck te...

Smoke Eater@rss.iscabbs.com — Mon, 06 Feb 2012 10:11:00 -0600

Which is exactly why he feels safe with the attitude of 'fuck testing'.  He
thinks 'even if it breaks, my job's safe, so so what?'

(John Public/89799) That sentiment is typical with trader types. They don't understa...

John Public@rss.iscabbs.com — Mon, 06 Feb 2012 09:56:00 -0600

That sentiment is typical with trader types. They don't understand the point
of QA.  The question is not the risk involved with the "fuck testing"
approach.  The question is, whose ass is on the line when the software
malfunctions.

Judging by his quip about scrambling to fix defects, he expects your ass to be
on the line, not his. :)

(Ozzy/89798) You've got to know your target user too. If you're writing medic...

Ozzy@rss.iscabbs.com — Mon, 06 Feb 2012 09:59:00 -0600

You've got to know your target user too. If you're writing medical software,
you might want to test more than your average programmer. Other fields, maybe
not so much.

(Smoke Eater/89797) Of course, you get the wrong type of management environment, and...

Smoke Eater@rss.iscabbs.com — Mon, 06 Feb 2012 09:55:00 -0600

Of course, you get the wrong type of management environment, and even a
programmer who learns from their mistakes will be thought of as a shithead for
making them in the first place.

(KalTorak/89796) Well... yeah.

KalTorak@rss.iscabbs.com — Mon, 06 Feb 2012 09:50:00 -0600

Well... yeah.
Know your environment.  If a busted release isn't catastrophic, that changes
the math on what sort of validation's wise.

(JL/89795) In a discussion last week with the client's guy in charge of all...

JL@rss.iscabbs.com — Mon, 06 Feb 2012 09:43:00 -0600


In a discussion last week with the client's guy in charge of all IT
development, including the financial app that we work on for them, he said:

"You know what I think - fuck testing. We don't need automated regression
testing or the money that it would take to implement it. So you move fast and
sometimes things break. So what? What's the longest that production has had
problems? A day or two? Big deal. You always fix it. Nobody expects perfection
when we drive you this hard and you release this fast. We can't afford to take
the time to put in a ridiculous test suite to test things that have worked just
fine for 10 years. I know it's chaos, but it works. We get way more
functionality built than the next guy, because we don't slow down to measure
every god damn thing and write exhaustive tests for every god damn thing. I
think we're doing it right. Fuck testing."

I admit, I kind of sat there dumbfounded. Not just because I'd never heard
it said quite like that, but because I kind of agreed with it. Kind of.

(Feoh/89794) Blarr. The House of Lords denied Alan Turing his pardon. Assha...

Feoh@rss.iscabbs.com — Mon, 06 Feb 2012 08:24:00 -0600

Blarr.  The House of Lords denied Alan Turing his pardon.  Asshats.
http://www.i-programmer.info/news/82-heritage/3735-widespread-celebrations-but-
no-pardon-for-turing.html

(John Public/89793) sftd. Danix, looking at the gsoap docs, all the examples call th...

John Public@rss.iscabbs.com — Mon, 06 Feb 2012 05:35:00 -0600

sftd. Danix, looking at the gsoap docs, all the examples call the global
init() before the soap_new() call, rather than the other way around. Also,
(and this is starting to feel familiar) the API has

    soap_done (struct soap*);
    soap_end  (struct soap*);
    soap_destroy  (struct soap*);
    soap_free (struct soap*);

each with subtly different semantics. Wow!

(John Public/89792) Snuffy> I agree that bitwise opera-TIONS are very useful. My rem...

John Public@rss.iscabbs.com — Mon, 06 Feb 2012 05:11:00 -0600

Snuffy> I agree that bitwise opera-TIONS are very useful. My remark was about
the opera-TORS. Developers go out of their way to avoid them, or bury bitwise
operators beneath more verbose macros.

Danix> I recall years ago tracing all sorts of serious issues (seg faults,
stack corruption, etc.) to gSOAP. We found that gSOAP was no longer
maintained, never appeared to be stable, and the developer who introduced it
was long gone. Our solution was to strip out gSOAP entirely. At first we
considered replacing gSOAP with XML-RPC or something, but then we realized the
data exchange was too simple to warrant a complicated solution like web
services -- the input was just a few input parameters, and the output was a
return code plus a blob of data. It was simpler than most CGIs!  Since there
weren't any complicated structures, there was no benefit to using to XML in
that case.  Have you found that gSOAP has improved?

(Danix/89791) Duh, I solved both this issue and another one I was getting furt...

Danix@rss.iscabbs.com — Sat, 04 Feb 2012 17:45:00 -0600

Duh, I solved both this issue and another one I was getting further down the
code.

Turns up that some functions being called between the soap_new() and
soap_free() calls are meant to be used only once, *and* that SSL works
differently for gSOAP. I had to switch this:

int do_something()
{
   struct soap *soap = soap_new();
   soap_ssl_init();
   soap_ssl_client_context(/* parameters to set up SSL */);
   // do everything
   soap_free(soap);
}

to this:

void initialize_stuff() {
   soap_master = soap_new();
   soap_ssl_init();
   soap_ssl_client_context(/* parameters to set up SSL */);
}

int do_somethin() {
   struct soap *soap = soap_copy(soap_master);
   // do something
   soap_free(soap);
}

and put struct soap *soap_master as a global var.

The second one was weirder. For some strange reason, a free()ing a string would
sometimes segfault with "free(): invalid next size (fast)". It seems that the
malloc'd space for that particular array was being overrun, so I just added 2
to the call (the size is dynamically taken from the returned string) and it
worked.

Ah, I have been romanticizing C a lot. I had forgotten about how easy it is to
crash a C program. :P

(Danix/89790) Hm, looks like the pointer isn't getting trashed, but something ...

Danix@rss.iscabbs.com — Sat, 04 Feb 2012 07:50:00 -0600

Hm, looks like the pointer isn't getting trashed, but something in that struct
might be.

ptr aqui: 8b43860d
ptr aqui (prefree): 8b43860d
*** glibc detected *** /opt/pdweb/bin/webseald: free(): invalid pointer:
0x08b127e8 ***
======= Backtrace: =========
/lib/libc.so.6[0xd300f1]
/lib/libc.so.6(cfree+0x90)[0xd33bc0]
/lib/libcrypto.so.6(CRYPTO_free+0x3a)[0x3aa180a]
/lib/libcrypto.so.6(BUF_MEM_free+0x4f)[0x3a26f3f]
/lib/libcrypto.so.6[0x3a442d2]
/lib/libcrypto.so.6[0x3a49e25]
/lib/libcrypto.so.6(ASN1_template_free+0x8b)[0x3a49feb]
/lib/libcrypto.so.6[0x3a49ef1]
/lib/libcrypto.so.6(ASN1_template_free+0x8b)[0x3a49feb]
/lib/libcrypto.so.6[0x3a49ef1]
/lib/libcrypto.so.6(ASN1_item_free+0x13)[0x3a4a033]
/lib/libcrypto.so.6(X509_free+0x27)[0x3a44ad7]
/lib/libssl.so.6(SSL_SESSION_free+0xd7)[0x916f47]
/usr/lib/libgsoapssl.so.0(soap_done+0x20d)[0x8a466d]
/opt/pdwebrte/daxtest_adk/example/libxauthn.so(vasco__valida_token+0x144)[0x8e5
02f]
/opt/pdwebrte/daxtest_adk/example/libxauthn.so(valida_token_canal+0x34)[0x8d972
4]
/opt/pdwebrte/daxtest_adk/example/libxauthn.so(xauthn_authenticate+0x410)[0x8d8
41d]
/opt/pdwebrte/daxtest_adk/example/libxauthn.so(pam_sm_authenticate+0x145)[0x8e5
335]
/usr/lib/libpdauthnv2.so(pd_pam_dispatch+0x104)[0x286894]
/usr/lib/libpdauthnv2.so(pd_pam_authenticate+0x31)[0x286ca1]
/usr/lib/libpdauthnv2.so(ivpam_authenticate+0xf6)[0x2859a6]
/usr/lib/libpdauthnv2.so(_ZN7PDAuthn12authenticateEPKcP18ivauthn_authn_infoR13P
DAuthnIDInfoRi+0x214)[0x283984]
/usr/lib/libpdauthzn.so(ivauthn_authenticate3+0xe5)[0xc699e5]
/opt/pdweb/bin/webseald[0x8092ecc]
/opt/pdweb/bin/webseald[0x808f1b0]
/opt/pdweb/bin/webseald[0x8180ccd]
/opt/pdweb/bin/webseald[0x8171af4]
/opt/pdweb/bin/webseald[0x8173cfd]
/opt/pdweb/bin/webseald[0x81752fa]
/opt/pdweb/bin/webseald[0x8176788]
/opt/pdweb/bin/webseald[0x815c216]
/opt/pdweb/bin/webseald[0x815cc6c]
/opt/pdweb/bin/webseald[0x815b516]
/opt/pdweb/bin/webseald[0x80db31a]
/opt/pdweb/bin/webseald(WsWorkerThread+0x85)[0x80dbb75]
/lib/libpthread.so.0[0x49149b]
/lib/libc.so.6(clone+0x5e)[0xd9842e]
======= Memory map: ========
00110000-00170000 r-xp 00000000 fd:00 4718851   
/opt/PolicyDirector/lib/libpdsvcutl.so
00170000-00178000 rwxp 0005f000 fd:00 4718851   
/opt/PolicyDirector/lib/libpdsvcutl.so
00178000-00179000 rwxp 00178000 00:00 0
00179000-001ee000 r-xp 00000000 fd:00 4718858   
/opt/PolicyDirector/lib/libpdz.6.1.1.so
001ee000-00204000 rwxp 00075000 fd:00 4718858   
/opt/PolicyDirector/lib/libpdz.6.1.1.so
00204000-00206000 rwxp 00204000 00:00 0
00206000-0021f000 r-xp 00000000 fd:00 4718855   
/opt/PolicyDirector/lib/libpdutil.so
0021f000-00223000 rwxp 00019000 fd:00 4718855   
/opt/PolicyDirector/lib/libpdutil.so
00223000-0026b000 r-xp 00000000 fd:00 4718826   
/opt/PolicyDirector/lib/libamaudutl.so
0026b000-00273000 rwxp 00048000 fd:00 4718826   
/opt/PolicyDirector/lib/libamaudutl.so
00273000-00274000 rwxp 00273000 00:00 0
00274000-0027b000 r-xp 00000000 fd:00 4752778   
/opt/pdwebrte/lib/libltpatoken.so
0027b000-0027e000 rwxp 00006000 fd:00 4752778   
/opt/pdwebrte/lib/libltpatoken.so
0027e000-00288000 r-xp 00000000 fd:00 4718845   
/opt/PolicyDirector/lib/libpdauthnv2.so
00288000-0028a000 rwxp 00009000 fd:00 4718845   
/opt/PolicyDirector/lib/libpdauthnv2.so
0028a000-0028b000 r-xp 00000000 fd:00 4752773   
/opt/pdwebrte/lib/libcdmfutils.so
0028b000-0028c000 rwxp 00001000 fd:00 4752773   
/opt/pdwebrte/lib/libcdmfutils.so
0028c000-0028d000 rwxp 0028c000 00:00 0
0028d000-00299000 r-xp 00000000 fd:00 4752770   
/opt/pdwebrte/lib/libamwzlib.so
00299000-0029d000 rwxp 0000b000 fd:00 4752770   
/opt/pdwebrte/lib/libamwzlib.so
0029d000-002a7000 r-xp 00000000 fd:00 7342161   
/usr/local/ibm/gsk7/lib/libgsk7iccs.so
002a7000-002a9000 rwxp 0000a000 fd:00 7342161   
/usr/local/ibm/gsk7/lib/libgsk7iccs.so
002a9000-002aa000 rwxp 002a9000 00:00 0
002aa000-002ae000 r-xp 00000000 fd:00 4718864   
/opt/PolicyDirector/lib/librspiutil.so
002ae000-002af000 rwxp 00004000 fd:00 4718864   
/opt/PolicyDirector/lib/librspiutil.so
002af000-002b1000 rwxp 002af000 00:00 0
002b1000-002d4000 r-xp 00000000 fd:00 4752763   
/opt/pdwebrte/lib/libamdsess.so
002d4000-002e7000 rwxp 00022000 fd:00 4752763   
/opt/pdwebrte/lib/libamdsess.so
002e7000-002e8000 rwxp 002e7000 00:00 0

Maybe gSOAP has a bug on the version I'm using?

(Tanj/89789) Use valgrind. Valgrind is your friend. If your environment doe...

Tanj@rss.iscabbs.com — Fri, 03 Feb 2012 09:06:00 -0600

Use valgrind.  Valgrind is your friend.  If your environment doesn't support
valgrind, figure out how to re-work it so that you can.

(Egregious/89788) Something is stomping on memory. To prove it, dump out the poin...

Egregious@rss.iscabbs.com — Fri, 03 Feb 2012 09:03:00 -0600

Something is stomping on memory.  To prove it, dump out the pointer after you
make it and just before you print it.

(Danix/89787) Ok. I've got some weird stuff going on. I'm using gSOAP, except ...

Danix@rss.iscabbs.com — Fri, 03 Feb 2012 08:57:00 -0600

Ok. I've got some weird stuff going on. I'm using gSOAP, except this is being
called from a TAM "authentication plugin". The code goes something like:

int do_this_auth(blahblah) {
   struct soap *soap = soap_new(); //this creates an "object". It's pure C.

   // call webservices, whatever.

   soap_free(soap);
}

The soap_free(soap); call works fine on the first time, the second ... but then
suddenly, and randomly, I'll get:

*** glibc detected *** /opt/pdweb/bin/webseald: free(): invalid pointer:
0x09accca0 ***

and everything shits itself with SIGABRT.

Huh?

(Egregious/89786) This is what sed is built for. I still don't appreciate the pow...

Egregious@rss.iscabbs.com — Fri, 03 Feb 2012 08:14:00 -0600


This is what sed is built for.  I still don't appreciate the power of sed,
but this is exactly its job.

sed -e 's/%40/@/'

(Stepwood/89785) With my shell script...

Stepwood@rss.iscabbs.com — Fri, 03 Feb 2012 08:04:00 -0600

With my shell script...

#!/bin/bash
grep -oE "[[:alnum:]]+%40[[:alnum:]+.[[:alnum:]]+"
/home/ubuntuuser/output/email.txt

This grabs the e-mail address from the email.txt file from a string that looks
like this...

http://link.com/0-5-10000406-138999419499616-redacted%40aol.com-9727d64937f46c6
4a91d77d745af2945

....and outputs it as redacted%40aol.com.

Is there a way, prior to outputting the e-mail address for the script to
replace the %40 with @? Maybe with sed or awk? So, that instead of outputting
redacted%40@aol.com it outputs redacted@aol.com?

(Do note: the .txt file cannot be altered)

(Feoh/89784) The Story of Mel, A Real Programmer is one of the best bits of h...

Feoh@rss.iscabbs.com — Fri, 03 Feb 2012 06:07:00 -0600

The Story of Mel, A Real Programmer is one of the best bits of hacker lore
ever.  I am glad I still own my first ed copy of the hacker's dictionary.  That
and the Xerox CP-V hack with Friar Tuck and Robin Hood which is also a favorite
:)
    Back in the mid-1970s, several of the system support staff at
     Motorola discovered a relatively simple way to crack system
     security on the Xerox CP-V timesharing system.  Through a simple
     programming strategy, it was possible for a user program to trick
     the system into running a portion of the program in `master mode'
     (supervisor state), in which memory protection does not apply.
     The program could then poke a large value into its `privilege
     level' byte (normally write-protected) and could then proceed to
     bypass all levels of security within the file-management system,
     patch the system monitor, and do numerous other interesting
     things.  In short, the barn door was wide open.
     Motorola quite properly reported this problem to Xerox via an
     official `level 1 SIDR' (a bug report with an intended urgency of
     `needs to be fixed yesterday').  Because the text of each SIDR
     was entered into a database that could be viewed by quite a
     number of people, Motorola followed the approved procedure: they
     simply reported the problem as `Security SIDR', and attached all
     of the necessary documentation, ways-to-reproduce, etc.
     The CP-V people at Xerox sat on their thumbs; they either didn't
     realize the severity of the problem, or didn't assign the
     necessary operating-system-staff resources to develop and
     distribute an official patch.
     Months passed.  The Motorola guys pestered their Xerox
     field-support rep, to no avail.  Finally they decided to take
     direct action, to demonstrate to Xerox management just how easily
     the system could be cracked and just how thoroughly the security
     safeguards could be subverted.
     They dug around in the operating-system listings and devised a
     thoroughly devilish set of patches.  These patches were then
     incorporated into a pair of programs called `Robin Hood' and
     `Friar Tuck'.  Robin Hood and Friar Tuck were designed to run as
     `ghost jobs' (daemons, in Unix terminology); they would use the
     existing loophole to subvert system security, install the
     necessary patches, and then keep an eye on one another's statuses
     in order to keep the system operator (in effect, the superuser)
     from aborting them.
     One fine day, the system operator on the main CP-V software
     development system in El Segundo was surprised by a number of
     unusual phenomena.  These included the following:
        * Tape drives would rewind and dismount their tapes in the
          middle of a job.
        * Disk drives would seek back and forth so rapidly that they
          would attempt to walk across the floor (see walking
          drives).
        * The card-punch output device would occasionally start up of
          itself and punch a lace card.  These would usually jam in
          the punch.
        * The console would print snide and insulting messages from
          Robin Hood to Friar Tuck, or vice versa.
        * The Xerox card reader had two output stackers; it could be
          instructed to stack into A, stack into B, or stack into A
          (unless a card was unreadable, in which case the bad card
          was placed into stacker B).  One of the patches installed by
          the ghosts added some code to the card-reader
          driver... after reading a card, it would flip over to the
          opposite stacker.  As a result, card decks would divide
          themselves in half when they were read, leaving the operator
          to recollate them manually.
     Naturally, the operator called in the operating-system
     developers.  They found the bandit ghost jobs running, and
     gunned them...  and were once again surprised.  When Robin Hood
     was gunned, the following sequence of events took place:
          !X id1
          id1: Friar Tuck... I am under attack!  Pray save me!
          id1: Off (aborted)
          id2: Fear not, friend Robin!  I shall rout the Sheriff
               of Nottingham's men!
          id1: Thank you, my good fellow!
     Each ghost-job would detect the fact that the other had been
     killed, and would start a new copy of the recently slain program
     within a few milliseconds.  The only way to kill both ghosts was
     to kill them simultaneously (very difficult) or to deliberately
     crash the system.
     Finally, the system programmers did the latter -- only to find
     that the bandits appeared once again when the system rebooted!
     It turned out that these two programs had patched the boot-time
     OS image (the kernel file, in Unix terms) and had added
     themselves to the list of programs that were to be started at
     boot time (this is similar to the way MS-DOS viruses propagate).
     The Robin Hood and Friar Tuck ghosts were finally eradicated when
     the system staff rebooted the system from a clean boot-tape and
     reinstalled the monitor.  Not long thereafter, Xerox released a
     patch for this problem.
     It is alleged that Xerox filed a complaint with Motorola's
     management about the merry-prankster actions of the two employees
     in question.  It is not recorded that any serious disciplinary
     action was taken against either of them.

(Cyanide/89783) Darphbobo> From way back.. I can think of a large number of proj...

Cyanide@rss.iscabbs.com — Fri, 03 Feb 2012 04:45:00 -0600

Darphbobo> From way back.. I can think of a large number of projects, but what
kind of programming are you wanting to do? If you want a large mixture, and
don't mind fixing bugs, go with something small like the Fedora ARM project.
It is a huge project, but a small subset of developers. Ubuntu ARM, is most
likely the same thing. but you basically can pick and choose what bug you
want to chase and you can work with the Fedora community and the software
project directly to help fix the issue.. When you find a project that really
interests you and you like the people, then you can become a package
maintainer, or just work with the project directly. But the trick is finding
something YOU want to do, and a project where you like the people you are
working with.

You need a way to "break in" IE you have a goal and you can help, versus
showing up on a dev list, and just saying, Oh I want to write a new feature,
which is akin to saying Oh I want to get married, but haven't ever been on
a date, muchless, not even familiar with the culture.

(Snuffy/89782) An example of places where bitwise operators are handy:

Snuffy@rss.iscabbs.com — Thu, 02 Feb 2012 16:01:00 -0600

An example of places where bitwise operators are handy:

Pack a value into the payload in little-endian format, with an offset
val = value to encode (unsigned)
stb = start bit (position of least significant bit, msb=63..........lsb=0)
enc = resulting 64-bit big-endian payload

% MATLAB:
% encoded signal:  shift left to payload byte boundary, mask bytes,
% re-order bytes to big-endian, re-assemble, shift right to final position

>> val = 174933; stb = 55;
>> bitshift( sum( bitshift( bitand( bitshift( uint64(val), mod(stb,8) ), ...
   bitshift( uint64(255), 8*[0:7] ) ), 8*[7:-2:-7] ), 'native' ), ...
   -8*(7-fix(stb/8)) )
ans =
   0080aa5501000000
>>

The first (value low-order) bit must be placed at bit 55 of the payload,
and then wrap to the next byte at byte boundary, continuing until all bits
of value are placed.

174933 = 0x2ab55 = 10.1010.1011.0101.0101 binary

bit 55-48:                                1.xxxx.xxx
bit 47-40:                      1.0101.010
bit 39-32:            0.1010.101
bit 31-24:  x.xxxx.xx1

 Byte 0   Byte 1   Byte 2   Byte 3   Byte 4   Byte 5   Byte 6   Byte 7
Tx-first ----------------------------------------------------- Tx-last
76543210 76543210 76543210 76543210 76543210 76543210 76543210 76543210
xxxxxxxx 1xxxxxxx 10101010 01010101 xxxxxxx1 xxxxxxxx xxxxxxxx xxxxxxxx
  0   0    8   0    A   A    5   5    0   1    0   0    0   0    0   0
MSB-----------------------------------------------------------------LSB
6      5 5      4 4      4 3      3 3      2 2      1 1      0 0      0
3      6 5      8 7      0 9      2 1      4 3      6 5      8 7      0

Thus:  0x0080AA5501000000 (packed signal value encoded in payload).

(Snuffy/89781) Understand ... I don't try to "out-clever" the compiler in other...

Snuffy@rss.iscabbs.com — Thu, 02 Feb 2012 14:33:00 -0600

Understand ... I don't try to "out-clever" the compiler in other more-general
contexts.

(Lochner/89780) The aversion to using bitwise operators directly is that the cod...

Lochner@rss.iscabbs.com — Thu, 02 Feb 2012 14:29:00 -0600


The aversion to using bitwise operators directly is that the code is hard to
read, and people don't remember what they do because we don't use them that
much.

I mean:

   x = x * 2;

The compiler turns this into:

   x <<= 1;

But I'd never write "x <<= 1" as code, because it's unnecessary.  The compiler
will make that optimization for me and in any case, it's harder to read than it
needs to be.

Likewise, I'd use a function or interface to set flags on a bitvector rather
than just place the operators in-line because it's just easier to read.

(Snuffy/89779) Feb 2, 2012 16:41 from Kena

Snuffy@rss.iscabbs.com — Thu, 02 Feb 2012 14:20:00 -0600

Feb 2, 2012 16:41 from Kena
I still remember the day when I had this flash: 6502 opcodes were *the same
thing* as data.
Needless to say, some self-modifying widget was written the same day.
[Programming And Software Engineering> msg #89774 (4 remaining)] Read cmd ->


To Kena's and Egregious's posts:
http://www.catb.org/jargon/html/story-of-mel.html

(Snuffy/89778) Lochner> I find it curious that languages offer bitwise operator...

Snuffy@rss.iscabbs.com — Thu, 02 Feb 2012 14:10:00 -0600

Lochner> I find it curious that languages offer bitwise operators, yet
everyone tries to avoid using them directly! Maybe the problem is the damn

I've been writing some CAN (controller area network) messaging code the last
few weeks.  Convert signals from "engineering" units (double) into various
packed formats, then loading the packed representations into the 64-bit CAN
packet payload, minding the big/little-endianness of the particular signal's
storage.   ---------->   bitwise operators are absolutely essential.

Actually, to be even more specific, I've been writing code that writes CAN
messaging code:  MATLAB script parses Vector DBC file output various source and
interfacing files.  The generated source has to run in two environments
(hardware ECU, big-endian, C++; simulation, little-endian, plain C).  The
hardware platform libraries are all C++, so it's not even "just use plain C
everywhere".

anyways, back to bit-bashing land ...

(Egregious/89777) I was looking through some 6502 code several months ago, and thi...

Egregious@rss.iscabbs.com — Thu, 02 Feb 2012 12:09:00 -0600


I was looking through some 6502 code several months ago, and this was
a pretty common pattern (paraphrasing):

$1000:    INCR $1004
$1002:    IF (50 <= 40)
$1005:    LONGJUMP $100a
$1007:    do stuff
$1009:    JUMP $1000
$100a:    rest of stuff

The increment command at line $1000 is modifying the code in line
$1002, where "40" is stored.  Enough times through the loop, and
the test in line $1002 will succeed.

Fun stuff, man.

(Feoh/89776) Core War is FUN!

Feoh@rss.iscabbs.com — Thu, 02 Feb 2012 11:51:00 -0600

Core War is FUN!
http://en.wikipedia.org/wiki/Core_War

(Kena/89775) Oh! Come to think of it, the "virus" bit, above, reminds me of ...

Kena@rss.iscabbs.com — Thu, 02 Feb 2012 11:42:00 -0600

Oh!  Come to think of it, the "virus" bit, above, reminds me of something that'
s been blamed for the entire *concept* of computer viruses: corewars.  (I'm not
sure *I* blame them, but I can see why one might think that.)  I'm on a
console, now, so I have no link, but I'm sure there's a Wikipedia page.

(Kena/89774) I still remember the day when I had this flash: 6502 opcodes wer...

Kena@rss.iscabbs.com — Thu, 02 Feb 2012 11:41:00 -0600

I still remember the day when I had this flash: 6502 opcodes were *the same
thing* as data.
Needless to say, some self-modifying widget was written the same day.

(Feoh/89773) I think self modifying code encompasses this just fine.

Feoh@rss.iscabbs.com — Thu, 02 Feb 2012 10:11:00 -0600

I think self modifying code encompasses this just fine.

Actually, so does 'metaprogramming' at least in some sense.

(Smoke Eater/89772) Viruses and malware?

Smoke Eater@rss.iscabbs.com — Thu, 02 Feb 2012 09:03:00 -0600

Viruses and malware?

(Egregious/89771) self-modifying code?

Egregious@rss.iscabbs.com — Thu, 02 Feb 2012 08:41:00 -0600

self-modifying code?

(JL/89770) On more of a programming note...

JL@rss.iscabbs.com — Thu, 02 Feb 2012 08:36:00 -0600


On more of a programming note...

While reading "Godel, Escher, Bach" I was intrigued by the idea of a program
being both the code and the data it consumes. (In the book, parallels are drawn
between AI and DNA/RNA).

An example given was a chess board. What if the layout of the board actually
determined the rules by which the pieces could move? So that on each move, the
rules would actually change because the layout had changed. Each subsequent
move would change both the "program" and the "data".

I want to see some examples of this, which I assume people have explored in
depth.

But I don't have a term for it, and my google-fu is failing.

Can anyone point me to applications of this concept?

(Faunus/89769) An on-checkin prettification thing would be kinda cool. Unless ...

Faunus@rss.iscabbs.com — Thu, 02 Feb 2012 08:26:00 -0600


An on-checkin prettification thing would be kinda cool.  Unless your
prettifier is not very reliable and might misunderstand something and
end up changing 80% of the code in a file due to misinterpretation of
some bit of syntax that changes indentation all over the place, and
now you've got a pain in the ass tracking changes because you've got
the whole damn file changed at one point, and it's automated so you
don't catch it till it's too late...

(Feoh/89768) JL> Precisely! I think being accountable for doing actual work ...

Feoh@rss.iscabbs.com — Thu, 02 Feb 2012 08:26:00 -0600

JL> Precisely!  I think being accountable for doing actual work is important,
but IMO being forced to present your findings is plenty of structure.  If
people are truly treating it as 'recess' then their presentation will be crap
and everyone will know it.  If your employees don't care that their colleagues
think of them as doing crap work, then you have a sick company and far bigger
problems to solve than time 'wasted' on R&D.

(JL/89767) Even better, don't force a coding style on developers, but conve...

JL@rss.iscabbs.com — Thu, 02 Feb 2012 08:04:00 -0600

Even better, don't force a coding style on developers, but convert it to your
style when they check in. Consistency without the battle.

JP>
>approached like any other venture or project.  Identify goals, commit to
>metrics, present a short-term plan, gather feedback, assemble a team, adapt,
>adjust, go for round two, etc.

I don't think real inspiration can emerge from a structured environment. That's
kind of the point.

I read an article about this recently, but I forget where it was. Basically,
smart people thrive and come up with new ideas when they get the chance to
"play". To explore paths that are interesting, rather than the ones that
promise a ROI. To pursue something in depth without knowing if it will have any
payoff, but do so because it challenges your mind to think in a new way.

Great things come from great minds in environments like this. Locking great
minds into controlled boxes with estimates, metrics, goals, and reviews, just
ruins the whole thing.

(Feoh/89766) That's just dumb. Pick a standard. Enforce adherence at the SC...

Feoh@rss.iscabbs.com — Thu, 02 Feb 2012 07:55:00 -0600

That's just dumb.  Pick a standard.  Enforce adherence at the SCM level if at
all possible (and if you're running something more moden than RCS it's possible
:)

Hell, find a coding standard on the net - pick any you like.  In the end
analysis it really doesn't matter.  Consistency is what matters.

(Egregious/89765) I was once at a company with a few CompSci professors on board a...

Egregious@rss.iscabbs.com — Thu, 02 Feb 2012 07:43:00 -0600

I was once at a company with a few CompSci professors on board and we
lost several weeks in tab fights.

(Feoh/89764) The One True coding style varies with each organization/team. A...

Feoh@rss.iscabbs.com — Thu, 02 Feb 2012 07:40:00 -0600

The One True coding style varies with each organization/team.  As long as
everyone coding a project agree it's all good IMO.

(John Public/89763) Feoh> I'm not saying the idea is bad or that it is universally a...

John Public@rss.iscabbs.com — Thu, 02 Feb 2012 07:05:00 -0600

Feoh> I'm not saying the idea is bad or that it is universally a waste of
time, or that I personally wouldn't *love* recess day.  But I know what I'd do
with my recess day, and I can assure you it wouldn't be for the benefit of the
company. ;) Over the past few decades, R&D has earned a bad name because big
companies like Bell decided to have selective vision: in a massive
bureaucracy, R&D is inefficient, so it gets shitcanned, but other massive
bureaucratic inefficiencies aren't as heavily scrutinized.  R&D just has to be
approached like any other venture or project.  Identify goals, commit to
metrics, present a short-term plan, gather feedback, assemble a team, adapt,
adjust, go for round two, etc.  R&D can still be creative and foster
innovation, and also have *some* structure to it. :)


Lochner> I find it curious that languages offer bitwise operators, yet
everyone tries to avoid using them directly! Maybe the problem is the damn
operators in the first place. Macros or inline functions? That debate has
finally been settled by the standards: inline functions. Tabs or spaces? I
think the general consensus is spaces only because nobody could agree how many
spaces a tab should represent.

(Faunus/89762) I'm familiar with the One True Brace Style, but I didn't think i...

Faunus@rss.iscabbs.com — Thu, 02 Feb 2012 07:15:00 -0600

I'm familiar with the One True Brace Style, but I didn't think it extended to
things like tabs vs spaces.  Still: spaces, duh.