#pentagram@irc.freenode.net logs for 16 Feb 2003 (GMT)

Archive Today Yesterday Tomorrow
Pentagram homepage


[00:10:14] <-- Dark-Star has left #pentagram ()
[00:58:47] --> Kirben has joined #pentagram
[00:58:48] --- ChanServ gives channel operator status to Kirben
[03:00:24] --> Cashman has joined #pentagram
[03:22:00] <-- Cashman has left IRC ()
[04:59:09] --> Coren_ has joined #pentagram
[08:40:31] <-- Kirben has left IRC (Read error: 104 (Connection reset by peer))
[08:40:32] --> Kirben has joined #pentagram
[10:04:50] --> Dark-Star has joined #pentagram
[10:14:57] --> Colourless has joined #Pentagram
[10:14:57] --- ChanServ gives channel operator status to Colourless
[10:48:24] --- DarkeZzz is now known as Darke
[10:58:48] --- Colourless is now known as Cless|Away
[12:41:22] <-- Kirben has left IRC ("System Meltdown")
[12:45:54] --- Cless|Away is now known as Colourless
[12:56:11] <-- Coren_ has left IRC ("using sirc version 2.211+KSIRC/1.2.4")
[13:28:35] --> wjp has joined #pentagram
[13:28:35] --- ChanServ gives channel operator status to wjp
[13:50:38] <Darke> I don't suppose either of you would have any suggestions as to a quick&simple testing suite that effectively just runs a executable with a batch of imput files and verifies the output is exactly what what is expected? I've found a couple, like dejagnu and a Test::Cmd module for perl, I was wondering if you just knew of something more... err... commonly used. *Grin*
[13:51:15] <Colourless> can't say I do. I know they exist, but I couldn't tell you anything about the :-)
[13:51:15] <Colourless> m
[13:51:37] <Colourless> i would imagine most of them are for $$$
[13:53:12] <Darke> What I 'really' want is something more complicated then what I have for fold, which is essentially 1) Run the program, pipe output to output.new file. 2) Diff output with output.verified file which was hand verified by me to be correct. 3) Do this for $(infinity) tests.
[13:54:24] <Darke> The way I'm doing now is a bit painful, slow and error prone if I've made a change that touches something in all the test cases in a predictable way. I'd rather not use the same method for the compiler, since it's a bit of a pain to add new tests.
[13:56:43] <Darke> The problem is lack of documentation for both those above. DejaGNU expects you to be adding test cases to gcc or gdb in it's docs, and Test::Cmd is a little sparse on examples as to how you do it exactly in their docs.
[13:58:19] <Colourless> i've got to go
[13:58:25] <Colourless> cya
[13:58:27] <Darke> Bye!
[13:58:29] <-- Colourless has left IRC ("casts invisibility")
[14:12:36] <wjp> how about bash + diff? :-)
[14:12:53] <wjp> ah, you already did that, I see :-)
[14:14:01] <Darke> Yes. Not the most elegant way I concluded. *grin*
[14:18:55] <Darke> Would also be nice if it was vauguely more 'cross platform' (aka, works relatively brainlessly) in windows and OSX. 'Cause I'd like it to work there at least, so I don't trip over compiler/environment specific b0rk3nness and not know about it (had this happen with conf/ remember?), and I don't want to go to all the effort of discovering each bash' installation... weirdness'. *grin*
[16:47:57] * Darke *loves* perl. *hugs perl*
[16:52:16] <wjp> :-)
[17:10:58] * Darke ummms... and errs... and denies having any relations with that interpreter! *affermiativenod* *innocentlook!*
[17:32:50] <wjp> lol
[17:33:01] <wjp> weird rabbit :-)
[17:34:23] <Darke> Hey! I'm not weird! Just... err... umm... differently challenged! *triumphantlook*
[17:34:44] <wjp> whatever you say :-)
[17:40:02] * Darke is allowed to be weird anyway. He's half asleep and slowly but surely getting somewhere writing a compiler. *grin*
[18:56:16] * wjp wows
[18:56:22] <wjp> actual mail to the pentagram ML
[18:57:59] * Darke looks, y'know, innocent.
[18:58:09] <wjp> Internal compiler error: Error reporting routines re-entered.
[18:58:12] <wjp> eeeek :-)
[18:58:32] * Darke blinkieblinks. Umm...
[18:59:05] <wjp> <wjp> screens/save_game.cpp: In member function `virtual void
[18:59:05] <wjp> <wjp> ua_save_game_screen::tick()':
[18:59:05] <wjp> <wjp> screens/save_game.cpp:388: jump to case label
[18:59:05] <wjp> <wjp> screens/save_game.cpp:378: crosses initialization of `ua_savegame sg'
[18:59:05] <wjp> <wjp>
[18:59:06] <wjp> <wjp> Internal compiler error: Error reporting routines re-entered.
[18:59:11] <wjp> to be more precise
[18:59:20] <wjp> no ICE when I remove that error
[18:59:37] <Darke> Umm... wow. *grin*
[19:00:25] <wjp> gcc 3.2-r1, btw, so this isn't the latest version
[19:01:31] <Darke> Admittedly, I'm sufficiently masochistic that I'd put the error back in and slowly reduce it to a test case to submit a bug report with. *grin*
[19:02:17] <wjp> it's the current uwadv CVS; enjoy ;-)
[19:04:11] <Darke> Hmm... no thanks. Don't have enough hdd space to do a full download/compile of that tree at the moment. Got pentagram crud everywhere I need to clean up first. *grin*
[19:06:49] <wjp> heh :-)
[19:07:29] <wjp> hmmm... changing the window title in SDL/gnome/sawfish really causes CPU usage to go up
[19:08:18] <wjp> the games pauses for about 1/4 second when it changes the window title
[19:08:32] <wjp> (and that time goes up a lot when there are more windows open)
[19:08:48] <wjp> sawfish also has 10% CPU usage instead of it's normal ~0.2%
[19:10:05] <Darke> Curious.
[19:11:23] <wjp> I looked into it some time ago, and apparently SDL does an XSync on changing the title
[19:36:51] * Darke yawns and decides to sleep, since the sun's coming up. *grin* Night!
[19:37:22] --- Darke is now known as DarkeZzz
[19:56:17] <-- wjp has left IRC ("gtg")
[22:17:54] --> Kirben has joined #pentagram
[22:17:54] --- ChanServ gives channel operator status to Kirben
[22:42:11] <-- DarkeZzz has left IRC (Read error: 110 (Connection timed out))