#pentagram@irc.freenode.net logs for 8 May 2005 (GMT)

Archive Today Yesterday Tomorrow
Pentagram homepage


[00:03:44] <wjp> there's still some confusion between various Gumps about the coordinatespace in which 'dims' is
[00:04:13] <wjp> I was expecting dims.w to be in gump coordinates, but in ScalerGump it's interpreted as parent coordinates
[00:04:40] <wjp> (that's currently preventing my 'keep ItemRelativeGump on-screen' algorithm from working properly :-) )
[00:05:15] <wjp> anyway, will finish it tomorrow. Sleep first :-)
[00:29:44] --> sbx has joined #pentagram
[01:30:04] --> Darke has joined #pentagram
[07:45:26] <-- Darke has left IRC ("Inficio-Infeci-Infectum")
[09:51:04] <-- oaQ^ has left IRC ("Reconnecting")
[09:51:08] --> oaQ^ has joined #pentagram
[09:51:47] <sbx> hi oaK^
[09:51:53] <sbx> oaQ^
[09:52:01] <oaQ^> hi sbx
[10:29:53] <wjp> Cless|notHere: still nothere?
[10:46:24] <Cless|notHere> i'm here actually
[10:46:29] --- Cless|notHere is now known as Colourless
[10:46:51] <sbx> :)
[10:48:56] <wjp> I think the use of the 'dims' in ScalerGump is inconsistent with the other gumps
[10:49:03] <wjp> oh, and hi :-)
[10:49:35] <Colourless> yes i noticed the comments
[10:49:54] <Colourless> getting dims on scaler gump is problematic..
[10:49:57] <Colourless> it works for itself
[10:51:21] <wjp> PointOnGump doesn't seem to work properly for ScalerGump currently
[10:52:11] <wjp> input is in parent coords; it is then transformed to gump coords, which in the case of ScalerGump means scaling the coordinates
[10:52:16] <Colourless> dmm
[10:52:23] <Colourless> then it needs to be fixed i guess
[10:52:25] <wjp> they're then checked against the 'dims' rectangle, which has unscaled width/height
[10:53:17] <wjp> in which order were you planning to apply scaler1 and scaler2?
[10:53:25] <wjp> first 1 and then 2? :-)
[10:54:59] <Colourless> yes
[10:55:19] <wjp> ok, so dims.w should be set to sw1
[10:55:41] <wjp> good; that means there's no need for impossible conditionals in the call to the Gump constructor :-)
[10:56:30] <Colourless> i think i use dims.h and dims.w to quick work out the size of the final size
[10:57:08] <wjp> I'll add width,height fields to ScalerGump for that
[10:57:50] <wjp> that way we can keep everything the same, just replacing dims.w/h by width/height
[10:58:36] <Colourless> need to just make sure that it all works properly
[10:58:58] <Colourless> could be unnoticed consequences
[10:59:12] * wjp nods
[10:59:18] <Colourless> but i guess if it draws properly and clicks are properly handled than it should be ok
[10:59:53] --> Chetic has joined #pentagram
[11:01:49] <wjp> seems to be working so far
[11:02:06] <wjp> and BarkGumps/AskGumps are created on-screen now
[11:07:49] --> Fingolfin has joined #pentagram
[11:07:49] --- ChanServ gives channel operator status to Fingolfin
[11:08:14] <wjp> hi Fingolfin
[11:08:55] <Fingolfin> hi wjp
[11:08:58] <wjp> did you already hear back from SF about what to do with nightly builds/snapshots?
[11:11:36] <Fingolfin> yeah
[11:11:53] <Fingolfin> they updated the site status for me ... ;-)
[11:12:06] * wjp takes a look
[11:12:17] <Fingolfin> https://sourceforge.net/docman/display_doc.php?group_id=1&docid=2352
[11:12:44] <wjp> the requirement has been rescinded until the service is made available... how considerate of them :-)
[11:14:41] <wjp> well, I'm glad I hadn't overlooked something; I couldn't find anything about it anywhere in the project admin section
[11:14:41] <Fingolfin> yeah, well, I noticed the requirment after the last admin-only sitewide news a few days ago
[11:14:46] <Fingolfin> the one about the stats system
[11:15:00] <wjp> yeah, I noticed it too
[11:15:11] <Fingolfin> so I talked to Jacob Moorman, a good friend of mine, and complained about the lack of warning and that I am now a rule breaker and how bad I feel, and ... :-)
[11:15:39] <Fingolfin> he pointed out that they "announced" the thing in the feb sitewide news... well they did insert a link to the relevant page, indeed, but I guess many other people like me missed it in there =)
[11:16:02] <Fingolfin> so I asked him that they should announce this again, but this time in a brief mail, so that it doesn't get lost in the middle of a dozen other items and legal text =)
[11:16:25] <Fingolfin> he said I was the first to complain, but they immediately updated the site status/docs to clarify the situation ... :-)
[11:17:20] <Fingolfin> which is good -- /. would have loved to peck on SF.net for this ("Evil SF.net changes site TOS over night" -- a typical /. headline, totally wrong ;-)
[11:18:13] <Kirben> Yes, that recent announcement was bad, considering the snapshot service still isn't available. Put a request for snapshot service for exult a few months back and still waiting...
[11:18:56] <wjp> Colourless: committed
[11:19:34] <Fingolfin> Kirben: yeah, but they didn't mean to be bad (SF.net is far from being evil, and they immediately jumped to clarify my concerns)
[11:20:04] <Colourless> dammit wjp...we're beginning to look like we are nearing a relesable state
[11:20:12] <Fingolfin> the problem of course is: I know lots of SF.net staff, some are my friends, so I know how they work and am sure they mean the right thing and will do the right thing... others will be afraid they are going to do the wrong thing, of course :-)
[11:20:29] <Fingolfin> if in doubt, SF.net will amend its rules, and not force stupid senseless rules on anybody just for the sake of it
[11:21:16] <Fingolfin> in fact they had soft limits on e.g. webspace even in the past, but they have been very very lax about enforcing them (and enforcing usually meant a mail like "Uhm, you are using 500 MB instead of 100 MB for you project web space, is that really necessary or could you maybe reduce it a bit?" ;-)
[11:21:20] <Kirben> I just hope their snapshot service is as good as web host, when it is finally released widely.
[11:21:21] <wjp> I wasn't really worried about SF doing 'bad things' :-)
[11:22:01] <wjp> Colourless: sorry about that ;-)
[11:22:38] <Colourless> yeah yeah yeah.... like you care
[11:23:50] <Fingolfin> Colourless, wjp: uhhh, that's bad. Hmmm. How about accidentally introducing a hard to debug bug which makes a release impossible for now?
[11:24:22] <wjp> oh, I already took care of that. I hardly tested that major gump change I made yesterday
[11:24:57] <Colourless> if (std::rand() == 0) do_something_stupid();
[11:24:58] <Colourless> ?
[11:25:11] <Kirben> Should I enable BUILTIN_DATA for Windows snapshots, is it safe enough ?
[11:25:34] <wjp> Kirben: sure :-)
[11:25:39] <wjp> it could use some testing :-)
[11:26:31] <Fingolfin> brilliant
[11:26:39] <wjp> do you know what to add to Makefile.mingw for it or should I add it?
[11:26:51] <wjp> Colourless: nah, that's too easy to debug; make it do_something_clever();
[11:26:58] <Fingolfin> Colourless: or, you could spend some time to port the pentragram midi code to exult, so that OSX users finally get good MIDI music in Exult ... :-)
[11:26:58] <Kirben> Yes, adding the define now.
[11:27:06] <wjp> (and then secretly make it do something stupid after all! *evil grin*)
[11:27:13] <Colourless> fing i've actually been working on that
[11:27:16] * Colourless gasps
[11:27:20] <Fingolfin> eeek
[11:27:54] <Fingolfin> I'll be back in a moment
[11:28:15] <-- Fingolfin has left IRC ("42")
[11:28:42] <wjp> hm, 'eeek' ? I hope it's nothing serious
[11:29:03] <Colourless> probably died because i've been writing code
[11:29:22] <Colourless> i wish exult code looked more like pentagrams
[11:29:38] <Colourless> the 'ogg' code hacked into exult makes my eyes hurt
[11:31:40] <Colourless> could probably port the entire pentagram audio system to exult and remove the exult mixer dependency (and just end up replacing it with directly neededing the ogg and vorbis libraries)
[11:31:44] <Colourless> so probably not a good idea
[11:34:08] --> Fingolfin has joined #pentagram
[11:34:08] --- ChanServ gives channel operator status to Fingolfin
[11:36:21] <wjp> wb
[11:43:13] <wjp> hmmmm
[11:43:23] <wjp> I wonder how I managed to miss this
[11:43:38] <wjp> the diary and other centered gumps are not quite centered anymore
[11:45:24] --- sbx is now known as sbx|afk
[12:05:10] <wjp> maybe we should confuse the people monitoring our website and do a website update today
[12:05:23] <Colourless> back date it :-)
[12:05:42] <wjp> "and the day after tomorrow, we will fix this and that" ? :-)
[12:08:52] <wjp> how peculiar
[12:09:14] <wjp> I'm cleaning my desk (which is peculiar in its own right ;-) ), and there are two Diablo II CDs lying here
[12:09:47] <wjp> hm, I must've "borrowed" my brother's once :-)
[12:09:53] <wjp> ah well :-)
[12:12:22] <Colourless> what sort of geek are you, cleaning out your desk
[12:12:52] <wjp> of course I'm just doing that to avoid doing actual work :-)
[12:13:01] <Colourless> ag
[12:13:03] <Colourless> ah
[12:13:09] <Colourless> then by all means :-)
[12:27:18] <Fingolfin> yeah yeah, I was doing the same earlier today...
[12:27:51] <Fingolfin> should be writing on my thesis. and see what I am doing now, cleaning up variable names in ScummVM, chatting on IRC and ICQ, and reading things about Lie groups <sigh>
[12:33:50] --> Darke has joined #pentagram
[12:54:36] <wjp> Kirben: is the builtin-data stuff working for you?
[12:55:30] <Kirben> I still need to alter Makefile more, distracted by ScummVM bug.
[13:02:49] <wjp> k :-)
[13:04:04] <wjp> hm, we should probably move the DATA_FILES list from data/module.mk to object.mk so that Makefile.mingw can use it too
[13:06:44] <wjp> committed that; I hope it doesn't conflict with anything you already did
[13:33:05] <-- Fingolfin has left IRC ("42")
[14:07:31] <wjp> heh, configure uses a rather interesting hack to determine the size of data types when cross compiling :-)
[14:08:10] <wjp> it tries to allocate an array of size (1 - 2 * !(sizeof(short) <= x))
[14:08:29] <wjp> this will give a compile-time error if the check sizeof(short) <= x is false
[14:08:47] <wjp> it then does a binary search over the possible values for sizeof(short) to determine the value :-)
[14:21:47] <Colourless> uh... surely there is an easier way
[14:30:20] <wjp> when cross-compiling?
[14:30:25] <Colourless> yes
[14:42:17] <wjp> I can't really think of one, to be honest
[15:08:12] <Kirben> Des this look right ? still fails to find data/data.h
[15:08:13] <Kirben> data/data.h: data2c.exe $(DATA_FILES)
[15:08:13] <Kirben> data2c.exe $(DATA_FILES)
[15:08:39] <wjp> the first argument to data2c.exe should be data/data.h
[15:08:51] <wjp> it probably overwrote one of the data files now, by the way
[15:10:22] <Kirben> odd, I still just get:
[15:10:24] <Kirben> filesys/data.cpp:24:23: data/data.h: No such file or directory
[15:10:54] <wjp> oh, you should probably add a line filesys/data.o: data/data.h somewhere in the Makefile
[15:11:30] <wjp> otherwise make doesn't know that filesys/data.o needs data/data.h
[15:25:04] <Kirben> Compiles fine now.
[15:27:17] <Kirben> So I no longer need to bundle any files in data ?
[15:27:28] <wjp> no
[15:28:21] <wjp> you can also remove the data= setting from the example config file
[15:29:11] * wjp hmmms.... if we wanted to, it should be possible to change things so that the windows installation instruction would read "Copy pentagram.exe to Ultima 8 directory. Run pentagram.exe"
[15:31:48] <wjp> don't forget to rm data/data.h in clean
[15:35:57] <Kirben> Done
[15:37:13] <wjp> are all messages related to missing data files and/or directories gone from stdout/stderr?
[15:37:32] <wjp> I think I disabled most of the if BUILTIN_DATA is set
[15:37:36] <wjp> s/the/them/
[15:38:56] <Kirben> No, logs look fine.
[15:41:05] <-- Kirben has left IRC ("Sleep")
[15:43:19] <-- Darke has left IRC (Read error: 110 (Connection timed out))
[16:19:38] <Colourless> you've modified a lot of files over the past few days
[16:20:26] <wjp> yes
[16:21:45] <Colourless> in FontManager.cpp do you think that there should be a default choice of where to find the SDL_ttf.h header
[16:22:04] <wjp> oh, did I put both choices in an #ifdef ?
[16:22:04] <Colourless> because at the moment if you don't have either HAVE_SDL_TTF_H or HAVE_SDL_SDL_TTF_H defined it just gives you a lot of errors
[16:22:17] <Colourless> #ifdef HAVE_SDL_TTF_H
[16:22:17] <Colourless> #include "SDL_ttf.h"
[16:22:17] <Colourless> #elif defined(HAVE_SDL_SDL_TTF_H)
[16:22:17] <Colourless> #include <SDL/SDL_ttf.h>
[16:22:17] <Colourless> #endif
[16:22:31] <wjp> I could've sworn I changed that afterwards
[16:22:41] <wjp> ah well, must've thought about it but forgot later
[16:22:53] <wjp> which do you think should be the default?
[16:23:08] <wjp> "SDL_ttf.h" ?
[16:23:11] <Colourless> yes
[16:23:34] <Colourless> or maybe <SDL_ttf.h> as a default but still keeping those two
[16:28:41] <wjp> hm
[16:28:56] <wjp> I'd probably go with "SDL_ttf.h" since chances are that it'll be in the same place as SDL.h
[16:29:38] <Colourless> ok
[16:32:03] <wjp> committed
[16:36:21] <Colourless> the arrays in data.h should probably be const
[16:36:43] <Colourless> atm they are just static
[16:40:02] <wjp> probably, yeah
[16:40:12] <wjp> committed that as well
[16:47:31] <Colourless> hmm
[16:47:50] <Colourless> this is failing for me:
[16:47:50] <Colourless> if (!configfileman->get("font/font/path", confontfile))
[16:49:08] <Colourless> because it would appear something in something is screwing up with line ends
[16:50:19] <wjp> hm, I think I did pass the as_text option to the in-memory-file correctly
[16:50:45] <Colourless> yeah looks like it. all my line ends are being stripped
[16:51:00] <Colourless> 0D and 0A...
[16:52:19] <wjp> what do the endlines look like in data.h?
[16:52:33] <wjp> 0x0D,0x0A properly?
[16:55:02] <Colourless> they are 0x0D and 0x0A
[16:55:15] <Colourless> ConvertTextBuffer() seems to be handling them properly...
[16:56:44] <Colourless> uh no it's not
[16:57:34] <Colourless> stripping the wrong one
[16:57:35] <Colourless> :-)
[16:58:11] <wjp> and readline skips \r... that'd do it :-)
[17:05:44] <Colourless> much better now :-)
[17:18:13] <-- sbx|afk has left IRC (Read error: 60 (Operation timed out))
[17:19:59] <Colourless> 09/05/2005 02:35 2,342,912 Pentagram.exe
[17:20:19] <Colourless> 06/05/2005 20:33 1,335,296 Exult.exe
[17:20:22] <Colourless> hmmm
[17:21:33] <Colourless> Pentagram is kind of huge... and the data files only take up a couple hundred kb
[17:40:31] <wjp> yes...
[17:41:07] <watt> ls -l u8.exe: -rw-r--r-- 1 root users 935440 Nov 18 1995 /home/data/ultima8/u8.exe
[17:41:40] <watt> hehe.... yeah, I know comparing 16-bit apples to 32-bit oranges
[17:42:00] <wjp> the scalers seem to take up a lot of space
[17:42:45] <wjp> my Bilinear* .o files are a megabyte in total
[17:42:50] <wjp> (stripped)
[17:44:35] <Colourless> could make them smaller at the expense of code speed
[17:48:32] <wjp> the x2y24 one is almost twice as large as the others
[17:48:50] <Colourless> yep that would make sense :-)
[17:49:15] <Colourless> since if you look at the code, it's about twice as large as the others
[17:50:01] <Colourless> we could probably cut out half of the scaler templates
[17:50:14] <Colourless> some of them are there for 'completeness' sake
[17:50:18] <wjp> bbl, dinner
[17:50:50] <Colourless> could #ifdef the un needed ones out for now, until we say actually need them :-)
[17:54:05] <Colourless> bed for me now i think
[17:54:08] <Colourless> night
[17:54:17] --- Colourless is now known as Cless|ZzzZzzZ
[19:50:37] <watt> oh.. the gimp-plugin isn't building... yeah, I should work on that thing soon.
[19:57:08] <wjp> I guess I should emerge a recent version of the gimp
[19:58:14] <wjp> ah, CPPFLAGS
[19:59:37] <watt> yeah. I'm not really worried about that right now.. the plugin won't be that important until it can save
[20:03:11] <wjp> hm, why search for gimptool-2.0 in configure when we need 2.2?
[20:11:27] --> sbx has joined #pentagram
[20:14:16] <wjp> do you mind if I remove the HAVE_GIMP define from configure.ac?
[20:14:47] <wjp> having it makes config.h change (triggering a full rebuild) if you toggle the gimp plugin with configure
[20:32:43] <wjp> ah, I see that gimp 2.2.6 installs gimptool-2.0... how consistent :-)
[20:52:17] --> Darke has joined #pentagram
[20:56:54] <sbx> hi Darke
[21:21:59] <-- Darke has left IRC (Read error: 104 (Connection reset by peer))
[21:24:34] --> Darke has joined #pentagram
[22:20:32] <watt> yeah..... It's a little wierd I guess
[23:54:35] --> Kirben has joined #pentagram
[23:54:35] --- ChanServ gives channel operator status to Kirben