#pentagram@irc.freenode.net logs for 2 Jun 2003 (GMT)

Archive Today Yesterday Tomorrow
Pentagram homepage


[04:33:07] --> Kirben2 has joined #pentagram
[04:33:08] <-- Kirben has left IRC (Read error: 54 (Connection reset by peer))
[04:45:39] --- Kirben2 is now known as Kirben
[07:14:07] --> wjp has joined #pentagram
[07:14:07] --- ChanServ gives channel operator status to wjp
[07:15:20] --- wjp is now known as wjp|work
[07:30:47] --- DarkeZzz is now known as Darke
[09:14:12] --> CashZzz has joined #pentagram
[09:14:16] --- CashZzz is now known as Cashman
[10:00:18] <-- Cashman has left IRC ()
[11:17:20] <Darke> Portability of POSIX directory opening/listing functions (opendir, etc). Are they?
[11:18:09] * Darke blinkblinks and unwinds the question. *ahem* Are the POSIX opening/listing functions portable across the platforms we support? Or are we going to have lots of exceptions?
[11:19:11] <Darke> I'm just writing a "get me all the filenames (that are files) in the directory I've provided" function. Nothing fancy or anything. *grin*
[11:30:30] <Darke> Oooooh! Maybe we should include a regex library and allow access to it though opcodes/intrinsics! That way we could go back to the days of yore, when you used to have to hunt 'n guess to find the *exact* keyword or phrase the designer wanted you to guess, and implement that sort of conversation interface!
[11:31:28] <Darke> It would satisfy the many, many people who simultaneously *loved* u6 *and* u8! (Of which I'm sure there are one, or maybe two... *grin*)
[11:49:06] --> Colourless has joined #Pentagram
[11:49:08] --- ChanServ gives channel operator status to Colourless
[11:49:30] <Colourless> hi
[11:52:31] <Darke> Hi!
[11:56:18] <Colourless> even though we don't see any problems yet, all the Item::I_getX and similar functions are supposed to be returning the X/Y/Z of the top most parent if the item is in a container.
[11:56:20] <wjp|work> Darke: I suggest looking at Exult's listfiles
[11:56:37] <wjp|work> oh, right; forgot to change that
[11:56:43] <wjp|work> file a bug report ;-)
[11:57:00] <Colourless> i was considering doing that :-)
[11:57:45] <Darke> wjp|work: Ta. I was just hunting around to see if we've done something like this already. *grin*
[11:58:26] <Darke> Ewwwwwwww! Now I wish I hadn't!
[11:58:58] <Colourless> hey, we did it for a reason :-)
[11:59:52] <Colourless> getting a directory list in some sort of platform dependant manner is something of a fantasy
[11:59:58] * Darke thinks all the world should be POSIX compliant. *nodnod* That would sort things out! *grin*
[12:01:52] <Colourless> well, that wouldn't be a problem for windows. i hear it is posix compliant
[12:02:04] <Colourless> wjp, i'm adding a new function GetLocationAbsolute() that will get the location of the top most parent. i'll update the intrinsics too
[12:03:47] <Darke> The problem is converting it to pentagram's new stucture of things and testing it. Shall I just cut&paste the entire file and require the porter to make and test the changes for their local system? Or just copy the few that we can support and go from there?
[12:05:44] <Colourless> wjp, one of I_getDirFromCoords and I_getDirToCoords is broken. But which one?
[12:06:10] <Colourless> i'm thinking it's I_getDirFromCoords
[12:07:57] <Colourless> item->y - y should have been y - item->y
[12:08:16] <Colourless> uh wrong way round
[12:08:27] <Colourless> y - item->y should have been item->y - y :-)
[12:14:37] * Darke spiiiiins Colourless around and around in circles! See if you can work out which way to subtract now! *grin*
[12:27:24] * Darke thinks he probably should go to sleep, if he's making lame jokes in the #errors of the various ports of the ListFiles function he can't fix. Wants to finish this stuff furst though. *grin*
[12:59:40] <Darke> Hmm... well. It works, 'cept it doesn't. U7's ones return the cannonical form /path/to/dir/file-we-want.foo rather then @wherever/file-we-want.foo which means other things break, since we're trying to access outside out sandbox. *grin*
[13:01:33] <Darke> Should be easily fixable, I hope. Just not tonight. *grin*
[13:07:35] <Darke> How are we handling violations of things like the filesystem sandbox? Terminate with extreme prejudice? ("Raw file system access detected, blah, blah, blah"... exit(-1)) Or something slightly more elegant?
[13:20:16] * Darke yawns and hops off to sleep. Night!
[13:20:16] --- Darke is now known as DarkeZzz
[13:30:28] --> Dark-Star has joined #pentagram
[13:31:12] <wjp|work> oh, right, I screwed up one of those intrinsics
[13:55:31] <-- Kirben has left IRC ("System Meltdown")
[15:13:20] <-- wjp|work has left IRC ("going home")
[15:40:22] --> wjp has joined #pentagram
[15:40:22] --- ChanServ gives channel operator status to wjp
[16:03:31] <Colourless> hmm, i think i broke somehing
[16:03:43] <Colourless> guards now do their patrols facing backwards :-)
[16:03:53] <wjp> :-)
[16:04:30] <Colourless> but i'm not going to try to find the problem now
[16:04:45] <Colourless> i would rather get all this stuff committed
[16:05:45] <wjp> go right ahead :-)
[16:05:47] <wjp> we can always fix bugs later :-)
[16:06:18] * Colourless looks at the list of modified files.... ugh
[16:06:49] <wjp> uh oh :-)
[16:07:10] <Colourless> oh yes :-)
[16:08:01] <Colourless> ow, 76 files modified. committing this is going to take all year :-)
[16:08:29] <wjp> and that's not even counting the time it'll take to fix the conflicts ;-)
[16:12:07] <Colourless> :-)
[16:17:31] <Colourless> i worked out what the original did to get the colours to use with BarkGump. It's just a simple arithmetic thing
[16:34:59] <Colourless> uh, that 76 didn't even include the 8 new files :-)
[16:35:58] <Colourless> done :-)
[16:36:00] <Colourless> have fun :-)
[16:36:12] <Colourless> and remember, I DON'T DO MAKEFILES :-)
[16:38:31] <wjp> right, let's see what happens :-)
[16:39:10] <wjp> hm, only two conflicts :-)
[16:39:47] <Colourless> only 2 :-)
[16:41:20] <wjp> yikes
[16:41:29] <Colourless> yes?
[16:41:47] <wjp> convert/Convert.h:66: invalid conversion from `IDataSource*' to `
[16:41:48] <wjp> IFileDataSource*'
[16:41:56] <wjp> (times ~100)
[16:42:50] <Colourless> probably needs the header included
[16:43:02] <wjp> which one's that?
[16:43:13] <Colourless> IDataSource.h but it's included
[16:43:34] <wjp> read1(ucfile) ?
[16:43:46] <wjp> why not ucfile->read1() ?
[16:43:55] <Colourless> where?
[16:44:05] <Colourless> uh, i don't know. I didn't write 'that' code :-)
[16:44:12] <wjp> Convert.h line 66 :-)
[16:46:00] <wjp> ah
[16:46:02] <wjp> Disasm.cpp line 88
[16:47:48] <wjp> why are all those functions defined on IFileDataSource's instead of IDataSource?
[16:47:48] <wjp> (I know it's not your code :-) )
[16:47:48] <Colourless> cause they are doing stuff to curOffset
[16:47:48] <Colourless> which is local to Disasm.cpp
[16:47:52] <wjp> yes, but why IFileDataSource?
[16:48:01] <Colourless> i don't know
[16:48:12] <Colourless> a lots of Darke's code just all used IFileDataSource
[16:48:33] <Colourless> i 'thought' i changed all of it over to IDataSource, but it looks like i forgot something :-)
[16:48:53] <wjp> just one problem... "if (!ucfile->good())"
[16:50:00] <Colourless> that line could probably be changes to something like if (ucfile->getPos() >= ucfile->getLen())
[16:50:43] <Colourless> it's not like good is being used there to do anything other than to check for eof
[16:51:36] <wjp> s/getLen/getSize/ :-)
[16:53:03] <Colourless> :-)
[16:53:06] <wjp> let's see...
[16:53:26] <Colourless> autocomplete didn't work in mirc, so i just had to 'guess' what the method was called :-)
[16:53:35] <wjp> /data/pentagram/pentagram/kernel/CoreApp.h:48: undefined reference to `CoreApp::application'
[16:54:08] <wjp> oooh! quotes :-)
[16:55:00] <Colourless> how could CoreApp::application be 'undefined'
[16:55:28] <wjp> you tell me :-)
[16:55:39] <Colourless> just try moving static CoreApp* application; to the start of the class
[16:55:46] <wjp> nah, it's simpler than that
[16:55:55] <wjp> CoreApp.o isn't being linked here :-)
[16:56:07] <Colourless> um why?
[16:56:37] <wjp> no idea
[16:56:56] <wjp> oh yikes
[16:56:56] <wjp> getting disasm to compile is nearly impossible right now :-)
[16:57:03] * wjp removes the disasm target
[16:57:33] <Colourless> what? it requires 'all' of pentagram?
[16:57:43] <wjp> ...and the fold target...
[16:57:47] <wjp> basically yes :-)
[16:58:25] <wjp> ...and shapeconv...
[16:59:33] <Colourless> oh, i know what i did :-)
[16:59:34] <wjp> something's including an old C++ header
[16:59:49] <Colourless> Console is where the problem is :-)
[17:00:01] <Colourless> it's not dependant on GUIApp :-)
[17:00:04] <Colourless> s/not/now/
[17:00:13] <wjp> oh, strstream in XMLTree.cpp
[17:00:16] <Colourless> and GUIApp is dependant onto everything :-)
[17:00:32] <Colourless> remove it
[17:00:35] <Colourless> unused
[17:00:44] <Colourless> something of mine from a few days ago
[17:01:07] * wjp tsk tsks and adds pent_include.h to ResizableGump.cpp :-)
[17:01:27] <wjp> (that _really_ broke things there :-) )
[17:01:32] <Colourless> ResizableGump... is not used
[17:02:26] <wjp> ok... it compiles :-)
[17:02:47] <wjp> but where did the framerate go?
[17:03:41] <Colourless> uh?
[17:03:58] <wjp> hm, or maybe that was just a particularly crowded screen
[17:05:04] <Colourless> where are you? :-)
[17:05:11] <wjp> avatar movement is showing the same weird jumping as before, I think
[17:05:16] <wjp> east tenebrae
[17:05:37] <wjp> framerate at the start is around 47-50
[17:05:55] <Colourless> yes, however, the problem is due to the avatar being moved between frames
[17:06:10] <wjp> bark gumps are nice :-)
[17:06:14] <Colourless> the camera and the item loose sync till the next frame
[17:07:10] <Colourless> it amazes me just how fast pentagram changing maps :-)
[17:07:24] <wjp> maybe we should add a disk-crunching delay? :-)
[17:07:44] <Colourless> well, 'should' have the fade out fade in
[17:07:57] <wjp> yes, that's missing
[17:08:09] <wjp> and we don't clean up temporary actors! ;-)
[17:08:36] <Colourless> and... i was having problems with Gumps, deleteing the was causing some level of memory corruption :-)
[17:08:48] <Colourless> so, BarkGumps are not deleted when they are closed :-)
[17:08:54] <Colourless> so we are leaking them too :-)
[17:09:14] <wjp> uh oh :-)
[17:10:27] <Colourless> Gump.cpp line 94, uncomment it
[17:25:12] <wjp> bbl, dinner
[17:47:13] <wjp> back
[17:48:03] <Colourless> wb
[17:48:25] <wjp> ok, that causes crashes here too
[17:48:53] <Colourless> the is good and bad i guess :-)
[17:49:01] <wjp> mostly good :-)
[17:49:22] <Colourless> yeah well figuring it out now :-)
[17:49:41] <Colourless> hmmm
[17:49:45] * Colourless thinks
[17:49:52] * Colourless thinks some more
[17:50:05] * Colourless goes and eximines code
[17:50:09] <wjp> ==2814== Invalid read of size 4
[17:50:09] <wjp> ==2814== at 0x807590E: Gump::AddChild(Gump*, bool) (gumps/Gump.cpp:339)
[17:50:09] <wjp> ==2814== by 0x807A6A2: Item::I_ask(unsigned char const*, unsigned) (world/Item.cpp:988)
[17:50:32] <wjp> (and then about 10 more in AddChild)
[17:51:03] <Colourless> i have a good idea what the problem might be
[17:52:05] <wjp> "app->getDestkopGump();" <-- ??
[17:52:22] <wjp> autocompletion has its disadvantages I guess :-)
[17:52:58] <Colourless> hehe :-)
[17:54:54] <Colourless> uh, i'm finding more than just a few bugs here
[17:55:13] <Colourless> then again, maybe not, but there is one problem
[17:57:28] <Colourless> which may have been it
[17:57:41] <Colourless> i'll commit the changes
[17:57:50] <wjp> focus_child seems to be invalid at times?
[17:58:00] <Colourless> yes, exactly :-)
[17:58:37] <wjp> specifically when deleting it in SetupLerp? :-)
[17:59:40] <Colourless> yes, because i wasn't chacking and updating to make sure that the gump being deleted wasn't the focus child
[18:01:38] <Colourless> ok, committed
[18:02:00] <Colourless> also removed Console's dependancies on GUIApp (now it only depends on CoreApp)
[18:04:14] <wjp> ah, but that pulls in everything too
[18:04:24] <Colourless> well, it shouldn't :-)
[18:04:32] <wjp> indeed
[18:06:19] <Colourless> the font useed by u8 is really bad :-)
[18:06:24] <wjp> yes :-)
[18:08:18] <wjp> hm, did you change animation speed? Things seem faster
[18:08:34] <Colourless> not as far as I know
[18:09:07] <Colourless> i 'did' make a number of process related changes
[18:09:53] <wjp> maybe some are interfering with my local process changes :-)
[18:10:10] <Colourless> quite possible
[18:11:20] <wjp> ah, another thing to do on map switch... kill gumps :-)
[18:11:35] <Colourless> yes
[18:11:50] <Colourless> need forgot about that
[18:12:03] <Colourless> s/need //
[18:12:36] <Colourless> a flag needs to be added to kill/keep gumps that should/shouldn't be killed
[18:12:59] <Colourless> then again, you could just kill all the gumps and recreate them
[18:13:22] <Colourless> but i don't like that idea
[18:13:30] <wjp> me neither
[18:14:18] <Colourless> probably just a flag such as item_dependant would be enough
[18:14:40] <Colourless> and then have a recursive call of KillItemDependantGumps (or some shorter name)
[18:14:51] <wjp> did the inventory gump stay open after a teleport?
[18:15:12] <Colourless> i'll find out
[18:15:17] <wjp> the big avatar status gump? the miniature status gump? :-)
[18:15:19] <Colourless> i don't think so though
[18:16:26] <Colourless> only mini status stays open
[18:19:08] <wjp> is it me or doesn't Shaana's axe really hit the head currently?
[18:20:00] <wjp> (haven't paid much attention to that bit lately, though... was mostly focused on the walking away part :-) )
[18:21:00] <Colourless> hard to say
[18:24:25] <Colourless> it looks about right comparing it to the original
[18:25:42] <wjp> at a framerate of 30Hz right?
[18:26:07] <Colourless> u8 for me runs 'slow'. however the animations seem to be running far too fast...
[18:26:25] <Colourless> might be a merge related problem somewhere
[18:26:40] <wjp> the original?
[18:26:54] <wjp> wouldn't surprise me if some of the timer magic U8 is doing doesn't quite work as expected in XP
[18:27:01] <Colourless> what?
[18:27:04] <Colourless> orignal what/
[18:27:10] <wjp> the original u8
[18:27:23] <Colourless> u8 run 'really' slow in winxp
[18:28:08] <Colourless> can't see any problems
[18:30:02] <Colourless> hmm, it seems as it's almost as if frameLimiting has been turned off in pentagarm
[18:30:30] <Colourless> no different between animationRate(33) and animationRate(1000)
[18:32:36] <Colourless> uh, yes there is :-)
[18:33:06] <Colourless> the shape animations are currently untimed :-)
[18:33:16] <Colourless> eveything else is still working as it should
[18:40:55] <Colourless> i have better be going
[18:40:59] <Colourless> cya
[18:40:59] <-- Colourless has left IRC ("casts invisibility")
[19:21:35] --> Dominus has joined #pentagram
[19:22:08] <Dominus> hey ho
[19:22:39] <Dominus> I tried for minutes to update makefile.mingw and for some strange reason it didn't work...
[19:22:55] <Dominus> until I remembered to actually save what I did in the editor...
[19:23:00] <Dominus> :-)
[19:39:05] <Dominus> it crashes for me with
[19:39:07] <Dominus> Searching for alternate console font... Not Found.
[19:39:08] <Dominus> Loading console font config: @data/fixedfont.cfg... Ok
[19:39:08] <Dominus> Error: Console font path not found! Unable to continue. Exiting.
[19:48:31] <wjp> heh, saving is useful ;-)
[19:49:21] <Dominus> any idea why it is bailing out?
[19:49:26] <wjp> what's in your fixedfont.cfg?
[19:49:38] <Dominus> it'S the one from cvs
[19:49:56] <wjp> is it in your data dir properly?
[19:50:02] <Dominus> yep
[19:50:15] <Dominus> Loading console font config: @data/fixedfont.cfg... Ok
[19:50:34] <wjp> sure the file isn't changed locally?
[19:50:52] <wjp> that error occurs when it doesn't contain the font/path item
[19:50:56] <wjp> uh, yikes
[19:51:05] <wjp> why did Darke create a new configuration object there
[19:51:06] <wjp> ?
[19:51:25] <Dominus> ?
[19:51:43] <wjp> not that it would cause that error, but it's unnecessary
[19:52:02] <wjp> our Configuration object support multiple config files, but he creates an entirely new Configuration object
[19:52:14] <wjp> s/support/supports/
[19:52:40] <wjp> are you really really really sure your fixedfont.cfg isn't corrupt? :-)
[19:52:47] <Dominus> my fixedfont.cfg looks like:
[19:52:48] <Dominus> <font>
[19:52:49] <Dominus> <path>
[19:52:49] <Dominus> @data/fixedfont.tga
[19:52:49] <Dominus> </path>
[19:52:49] <Dominus> <height>
[19:52:51] <Dominus> 8
[19:52:53] <Dominus> </height>
[19:52:55] <Dominus> <width>
[19:52:57] <Dominus> 8
[19:52:59] <Dominus> </width>
[19:53:01] <Dominus> </font>
[19:53:08] <wjp> yes, unfortunately that's correct
[19:53:15] <wjp> did you try a clean make?
[19:53:16] <Dominus> and makfile.mingw just copies the one from cvs over everytime
[19:53:26] <Dominus> yes
[19:53:49] <Dominus> actually I do a clean one everytime now with pentagram
[19:54:08] <wjp> doesn't the 'normal' build system work for you?
[19:55:01] <wjp> anyway, I'm rather stumped
[19:55:01] <Dominus> oh it does but at this stage it doesn't take too long for a full recompile
[19:55:17] <wjp> hm, maybe endline style
[19:55:34] <wjp> Colourless recently switched the way Configuration reads text files
[19:55:40] <Dominus> well, it used to work before today
[19:55:55] <Dominus> so it could very well be that
[19:56:05] <wjp> he committed that today :-)
[19:56:12] * wjp takes a peek
[19:56:23] <Dominus> yep that'S it
[19:56:40] <Dominus> I just converted the cfg to unix style and it works
[19:57:03] <wjp> weird, it looks like it should work for dos-style endlines
[19:58:04] <Dominus> hey, left click on fish, then right click -> crash
[19:58:36] <wjp> no crash here
[19:58:44] <wjp> but I'll take closer look in a minute
[19:58:57] <Dominus> obviously looks are deceiving (-><wjp> weird, it looks like it should work for dos-style endlines)
[19:59:30] <wjp> in fact, when I convert my fixedfont.cfg to dos style it works
[19:59:37] <Dominus> ah, it happens when you click right very fast after left click
[19:59:44] <wjp> while the text is still there?
[19:59:57] <wjp> ah, yes
[20:01:09] <wjp> we don't check if the item is still there it seems
[20:01:10] <Dominus> Proton (my editor of choice) offers either unix, windows or mac style line breaks
[20:01:17] <wjp> mac style should break
[20:01:30] <Dominus> yep, I was about to say that :-)
[20:01:41] <wjp> were you using mac style?
[20:01:54] <Dominus> oops, I confused it
[20:01:59] <Dominus> mac style works
[20:02:19] <Dominus> windows breaks
[20:03:44] <wjp> eh?
[20:04:09] <Dominus> mac style actually allows pentagram to start
[20:04:23] <Dominus> but windows doesn't
[20:04:48] <wjp> oh, wait; mac style does work, yes; readline() just doesn't behave as expected
[20:05:40] <wjp> hm, I wonder what exactly happens when a file is openend in text mode
[20:07:06] <wjp> aha... the file isn't opened in text mode at all
[20:07:24] <wjp> try changing
[20:07:25] <wjp> if(fontconfig->readConfigFile(confontcfg, "font"))
[20:07:26] <wjp> into
[20:07:35] <wjp> ack, no, nvm :-)
[20:07:44] <wjp> confused :-)
[20:08:32] <wjp> ok, I want to see if this works:
[20:08:44] <wjp> in IDataSource.h, try changing line 101:
[20:08:48] <wjp> else if (character == '\n') break;
[20:08:48] <wjp> into
[20:08:54] <wjp> else if (character == '\n') continue;
[20:09:34] <Dominus> only that?
[20:10:48] <Dominus> still crashes
[20:13:24] <wjp> after a clean build?
[20:15:26] <Dominus> I just removed the o files in /filesys
[20:16:16] <wjp> wrong ones :-)
[20:16:27] <wjp> you need the ones in conf/
[20:16:45] <Dominus> oops
[20:17:29] <Dominus> but now that I think more about it , it makes sense somehow :-)
[20:17:46] <Dominus> still crashes
[20:18:17] <wjp> same error?
[20:18:37] <Dominus> yep
[20:18:56] <wjp> ok... let's see...
[20:18:56] <Dominus> damn this difference between english and german!!!!
[20:19:10] <wjp> any particular difference? :-)
[20:19:24] <Dominus> I almost bid on something on ebay.uk for 3.600 GBP
[20:19:29] <wjp> ouch
[20:19:33] <wjp> that's rather a lot :-)
[20:19:38] <Dominus> I put in 36,00 instead of 36.00
[20:20:13] <wjp> of course it probably wouldn't go all the way to 3600 :-)
[20:20:25] <Dominus> no, but still
[20:20:37] <Dominus> rather annoying...
[20:20:40] <wjp> anyway, let's add some debugging stuff:
[20:20:51] <wjp> in XMLTree::readConfigFile, right before 'delete f;'
[20:20:53] <wjp> add:
[20:21:06] <wjp> perr << "sbuf = " << sbuf << std::endl;
[20:22:41] <wjp> (should be about line 67)
[20:23:44] <Dominus> }
[20:23:44] <Dominus> perr << "sbuf = " << sbuf << std::endl;
[20:23:44] <Dominus> delete f;
[20:24:02] <wjp> yes
[20:24:32] <Dominus> oh, wait, it could be that I had reverted idatasource.h already to its original state...
[20:24:38] * Dominus slaps himself
[20:24:53] * Dominus thinks multitasking should be turned off for now
[20:26:50] <Dominus> well it crashes anyway
[20:26:58] <Dominus> in stderr I now have
[20:27:06] <Dominus> virtual path "@home": .
[20:27:06] <Dominus> virtual path "@data": data
[20:27:06] <Dominus> sbuf = <te eol>ddp\a\acf>
[20:27:06] <Dominus> sbuf = <aaa <ts< a>e<ori>ikam /k<lf_t<macf>
[20:27:06] <Dominus> virtual path "@u8": .
[20:27:06] <Dominus> virtual path "@work": ./u8-work
[20:27:08] <Dominus> sbuf = <t/thhg<>wh<o>
[20:27:16] <wjp> sorry?
[20:28:42] <Dominus> same as before only in stderr I have these additional lines
[20:28:43] <Dominus> virtual path "@data": data
[20:28:44] <Dominus> sbuf = <te eol>ddp\a\acf>
[20:28:44] <Dominus> sbuf = <aaa <ts< a>e<ori>ikam /k<lf_t<macf>
[20:29:27] <wjp> cute
[20:29:42] <wjp> after "sbuf = " you should get the contents of your config files
[20:29:58] <wjp> sbuf = <config> <general> <data> /data/pentagram/pentagram/data </data> </general> <u8> <path> /data/u8 </path> </u8> </paths> <video>
[20:29:58] <wjp> <width> 640 </width> <height> 480 </height> <bpp> 32 </bpp> <fullscreen> no </fullscreen> </video> <theme> default.theme </theme></config>
[20:30:03] <Dominus> he he
[20:30:03] <wjp> sbuf = <font> <path> @data/fixedfont.tga </path> <height> </height> <width> 8 </width></font>
[20:30:05] <wjp> that's what I guet
[20:30:08] <wjp> s/guet/get/
[20:31:03] <Dominus> he he, if I save with unix ending I get this too
[20:31:22] <Dominus> instead of virtual path "@work": ./u8-work
[20:31:24] <Dominus> sbuf = <t/thhg<>wh<o>
[20:31:42] <Dominus> I get virtual path "@work": ./u8-work
[20:31:43] <Dominus> sbuf = <font> <path> @data/fixedfont.tga </path> <height> 8 </height> <width> 8 </width></font>
[20:31:53] <Dominus> but only with unix style linebreaks
[20:32:09] <wjp> uh
[20:32:19] <wjp> right
[20:32:24] <wjp> <-- confused
[20:33:14] <Dominus> well, at least now I understand that it crashes :-)
[20:33:23] <wjp> you can revert IDataSource.h, btw
[20:33:29] <wjp> the problem is somewhere else :-)
[20:33:46] <Dominus> when I compare <t/thhg<>wh<o> to what should be there I'd crash too
[20:34:01] <wjp> :-)
[20:34:16] <wjp> the weird part is that it's still letters and <, /, >
[20:34:42] <wjp> like it's taking every 10th character or something :-)
[20:35:01] <Dominus> yeah
[20:35:04] <wjp> did you recently upgrade anything?
[20:35:47] <Dominus> huh?
[20:35:51] <Dominus> no
[20:35:56] <wjp> any libraries? compiler?
[20:36:02] <wjp> change your optimization settings?
[20:36:06] <Dominus> nope same as before
[20:37:49] * wjp hmmms
[20:38:05] <wjp> try compiling without optimization
[20:38:23] <wjp> (so remove the -O2 on the CXXFLAGS=-O2 line)
[20:39:26] <Dominus> cflags too?
[20:39:33] <wjp> no need
[20:41:04] <Dominus> no change
[20:42:08] <wjp> this must be one of the weirdest bugs I've seen
[20:42:34] <wjp> I suppose things work if you change XMLTree.cpp line 59 to
[20:42:41] <wjp> IDataSource *f = FileSystem::get_instance()->ReadFile(fname, false);
[20:42:42] <wjp> ?
[20:42:46] <wjp> (so s/true/false/ there)
[20:44:10] <Dominus> yes
[20:45:24] <wjp> this kind of sounds like a bug in your C library
[20:45:50] <wjp> character-based access to a text stream is apparently broken
[20:46:03] <wjp> (or your C++ library)
[20:46:21] <Dominus> hmm
[20:46:51] <wjp> feel like running some test programs to narrow it down? :-)
[20:47:18] <Dominus> feel like it yourself?
[20:47:35] <Dominus> then I'd say yes :-)
[20:52:53] <wjp> http://www.math.leidenuniv.nl/~wpalenst/testchar.cc
[20:52:55] <wjp> compile with
[20:53:00] <wjp> g++ -o testchar testchar.cc
[20:53:09] <wjp> run it in a directory which contains fixedfont.cfg
[20:53:18] <wjp> or -o testchar.exe, probably :-)
[20:56:03] <Dominus> hmm
[20:56:33] <wjp> don't tell me it outputs two correct copies of fixedfont.cfg? :-)
[20:56:35] <Dominus> fopen and fstream both show me correct entries for the cfg
[20:56:44] <Dominus> :-)
[20:57:35] <wjp> I wonder if it tries to open the wrong file
[20:57:55] <Dominus> huh?
[20:58:16] <Dominus> some things I noticed that might help in finding:
[20:58:36] <Dominus> after
[20:58:37] <Dominus> virtual path "@home": .
[20:58:38] <Dominus> virtual path "@data": data
[20:59:30] <Dominus> in stderr it shows garbage as well (only after IDataSource *f = FileSystem::get_instance()->ReadFile(fname, false); it showed correct stuff as well)
[21:00:07] <Dominus> but pentagram still runs after only correcting linebreaks for fixedfont.cfg
[21:00:37] <Dominus> in stderr it shows garbage as well -> I mean the sbuf = stuff
[21:01:10] <wjp> FileSystem.cpp, line 116 (right after switch_slashes(name); )
[21:01:17] <wjp> try adding
[21:01:24] <wjp> perr << "name = " << name << std::endl;
[21:01:56] <wjp> although it probably will work ok
[21:02:14] <wjp> since I don't really see how changing text/non-text will affect this
[21:03:32] <wjp> new version of testchar.cc too if you want to try it :-)
[21:03:37] <Dominus> same, only stderr now says
[21:03:39] <Dominus> virtual path "@home": .
[21:03:39] <Dominus> virtual path "@data": data
[21:03:39] <Dominus> name = data\pentagram.cfg
[21:03:39] <Dominus> sbuf = <te eol>ddp\a\acf>
[21:03:39] <Dominus> name = .\pentagram.cfg
[21:03:40] <Dominus> sbuf = <aaa <ts< a>e<ori>ikam /k<lf_t<macf>
[21:03:42] <Dominus> virtual path "@u8": .
[21:03:44] <Dominus> virtual path "@work": ./u8-work
[21:03:46] <Dominus> name = data\fixedfont.cfg
[21:03:48] <Dominus> sbuf = <t/thhg<>wh<o>
[21:04:10] <wjp> do data\pentagram.cfg and .\pentagram.cfg exist?
[21:04:46] <Dominus> oops, yes :-)
[21:04:55] <wjp> oops?
[21:05:53] <Dominus> the one in data has not everything in it
[21:06:38] <Dominus> new version of testchar.cc still shows correct stuff
[21:07:35] <wjp> I wonder what kind of a simple/stupid thing this is going to turn out to be :-)
[21:09:02] <Dominus> why is it correctly using pentagram.cfg but not fixedfont.cfg
[21:09:15] <Dominus> it's gotta be something really simple
[21:09:50] <Dominus> I suggest we just don't worry about it anymore and think about it tomorrow
[21:10:20] <wjp> well, it's showing garbage for pentagram.cfg too
[21:11:37] <wjp> are you sure it isn't just working with default values?
[21:11:55] <Dominus> hmm, just thought that myself
[21:13:23] <Dominus> *cough*
[21:13:40] <Dominus> it doesn't read it correctly as well
[21:13:56] <wjp> one more debugging info line:
[21:14:08] <wjp> IDataSource.h, line 103, before "str+= character;", add:
[21:14:15] <wjp> perr << character;
[21:14:41] <wjp> (recompile at least conf/)
[21:16:17] <Dominus> name = data\pentagram.cfg
[21:16:17] <Dominus> <tkpk/llm//>lnn<le i>/ngsbuf = <tkpk/llm//>lnn<le i>/ng
[21:16:17] <Dominus> name = .\pentagram.cfg
[21:16:17] <Dominus> <tkpk/llm//>lnn<le i>/ngsbuf = <tkpk/llm//>lnn<le i>/ng
[21:16:17] <Dominus> virtual path "@u8": .
[21:16:18] <Dominus> virtual path "@work": ./u8-work
[21:16:20] <Dominus> name = data\fixedfont.cfg
[21:16:22] <Dominus> <t/thhg<>wh<o> sbuf = <t/thhg<>wh<o>
[21:17:23] <wjp> unbelievable
[21:17:38] <wjp> (well, this was pretty much expected, but still :-) )
[21:17:48] <wjp> try changing that added line to:
[21:17:56] <wjp> perr << character << std::endl;
[21:17:59] <DarkeZzz> We're not doing something silly like escaping \? characters into their binary format, are we?
[21:18:01] <wjp> (and don't copy paste the output in here after that :-) )
[21:18:15] <wjp> \? ?
[21:18:37] <DarkeZzz> \a \n etc.
[21:18:47] <DarkeZzz> Just thinking of windows path names. *grin*
[21:19:15] <wjp> morning, btw :-)
[21:19:19] <Dominus> oh, please let me copy paste
[21:19:25] <Dominus> :-)
[21:19:35] <wjp> does it show the garbage line-by-line?
[21:19:39] <wjp> or anything unexpected?
[21:19:51] <Dominus> garbage line-by-line
[21:20:03] <wjp> WHY?!? *cough*
[21:20:24] <wjp> uh oh
[21:20:33] <Dominus> (except for pentgarm.cfg, that I changed to unix line-break now)
[21:20:35] * wjp lightbulbs
[21:20:51] * Dominus fetches his sun glasses
[21:20:53] <wjp> as I said, it's something quite simple/stupid :-)
[21:21:54] <DarkeZzz> BTW, I kept the 'font' stuff as a seperate config object, since I wasn't sure how we were going to handle loading/saving of config files and such. I'd rather not have us save the font stuff into the global config. *grin*
[21:22:14] <wjp> DarkeZzz: each config file can only have one root
[21:22:27] <wjp> it saves new settings into the config file with the correct root that was loaded last
[21:22:49] * DarkeZzz aieees! The code! The code it's *full* of conflicts! *whimperhide!*
[21:23:02] <wjp> Dominus: new testchar.cc
[21:24:06] * wjp sighs
[21:24:12] <wjp> I should've picked that up hours ago
[21:24:14] <Dominus> ok, you got it in fstream
[21:24:39] <Dominus> so what is it, Doctor?
[21:24:40] <wjp> seeking in text files doesn't work properly if it's translating endlines
[21:25:35] <wjp> mac and unix work ok because endlines are just one character
[21:26:00] <Dominus> testchar: fopen: correct, fstream:garbage (to clarify my above comment)
[21:26:08] <wjp> Dominus: yeah, I know :-)
[21:26:53] <wjp> I'll fix it later
[21:27:03] <wjp> for now you can just load files in binary mode
[21:27:06] <Dominus> okay
[21:27:10] <wjp> (that was replacing the 'true' with 'false')
[21:27:59] <Dominus> later=tonight/tomorrow or later=at some not so distant point in the future ?
[21:28:15] <wjp> I'll see :-)
[21:28:16] <Dominus> I'm okay either way now :-)
[21:28:30] <wjp> I think I'll go submit some bug reports :-)
[21:29:10] <Dominus> he he
[21:29:30] <Dominus> glad you've found it
[21:43:33] <DarkeZzz> Hmm... getting anything other then the main app to compile is a bit... tricky at the moment. *grin*
[21:45:32] * DarkeZzz hmms... and looks like he's going to be late for work this morning. Ah well, he'll just blame Colourless, and they'll be after the dragon with torches and pitchforks. *grin*
[21:46:27] <DarkeZzz> Actually, knowing QUT's illogical stupidities with reguarding employment contracts 'n stuff, it's entirely possible they'll hunt Colourless down, only to sacrifice virgin maidens to him, or something. You never know with this set of beurocratic loonies...
[21:47:44] <Dominus> I thought onyl Wolfram&Heart would do that...
[21:49:11] <Dominus> it's nice to walk through maps with current pentagram
[21:49:19] <Dominus> I mean "walk"...
[21:50:07] * DarkeZzz snickers.
[21:50:35] <DarkeZzz> Of course, they could just tar 'n feather Colourless, but no doubt after that order got through the chain of command, we'd just end up with a compressed micro-dragon.
[21:50:51] <Dominus> oh, no
[21:50:58] <Dominus> micro-dragon is someone else
[21:51:37] <DarkeZzz> Cool. Got it to compile and work, finally. *grin*
[21:52:25] <wjp> I should go; 'night
[21:52:26] <Dominus> I guess the time the avatar'S gump (text selection) is active will be prolonged in the future ...
[21:52:34] <Dominus> bye
[21:52:36] <wjp> Dominus: no, that'll be totally changed
[21:52:46] <DarkeZzz> Bye!
[21:52:48] <Dominus> I see
[21:52:51] <wjp> it'll stay until you select something :-)
[21:53:06] <Dominus> he he
[21:53:13] <wjp> anyway, bye :-)
[21:53:17] <-- wjp has left IRC ("Zzzz...")
[21:53:33] <Dominus> I'm not complaining about something like this at the current state of pentagram
[21:54:33] * DarkeZzz goes too. He needs to blame stuff on dragons. *grin* Bye!
[21:54:55] <Dominus> bye bye
[22:23:48] <-- Dominus has left IRC ("enough for now")
[22:57:21] <-- Dark-Star has left IRC ()
[23:24:44] --> Kirben has joined #pentagram
[23:24:44] --- ChanServ gives channel operator status to Kirben