Config
Log for #openttdcoop.devzone on 12th August 2010:
Times are UTC Toggle Colours
00:04:20  *** KenjiE20 has quit IRC
06:00:12  *** frosch123 has joined #openttdcoop.devzone
06:29:26  <Brot6> NFORenum - Revision 464:70ad56d819d5: Fix: bogus warnings when defining a range strings and then ... (Rubidium) @ http://dev.openttdcoop.org/projects/nforenum/repository/revisions/70ad56d819d5
06:47:00  <Brot6> NFORenum - Revision 465:bd682578a01f: Fix: make dependency failed a bit and regenerating all dep ... (Rubidium) @ http://dev.openttdcoop.org/projects/nforenum/repository/revisions/bd682578a01f
06:51:41  <Brot6> GRFCodec - Revision 223:3f08131bd6a1: Fix: make dependency failed a bit and regenerating all dep ... (Rubidium) @ http://dev.openttdcoop.org/projects/grfcodec/repository/revisions/3f08131bd6a1
08:03:27  *** thgergo has joined #openttdcoop.devzone
08:11:23  *** thgergo has quit IRC
08:11:40  *** thgergo has joined #openttdcoop.devzone
08:13:36  *** ODM has joined #openttdcoop.devzone
09:56:36  *** Seberoth has joined #openttdcoop.devzone
10:41:09  *** KenjiE20 has joined #openttdcoop.devzone
12:23:22  *** Seberoth has quit IRC
12:27:15  *** Seberoth has joined #openttdcoop.devzone
12:56:23  *** Yexo has quit IRC
12:56:42  *** Yexo has joined #openttdcoop.devzone
13:29:41  *** Seberoth has quit IRC
13:31:53  *** Seberoth has joined #openttdcoop.devzone
15:11:28  <Brot6> grfcodec: update from r222 to r223 done - http://bundles.openttdcoop.org/grfcodec/nightlies/r223
15:11:36  <Rubidium> early bird!
15:12:09  <Ammler> yeah, I should know, why my diffdiff doesn't work :-)
15:13:27  <Brot6> 2cctrainset: rebuild of r573 done (6 errors) - http://bundles.openttdcoop.org/2cctrainset/nightlies/r573/log
15:20:08  *** Doorslammer has joined #openttdcoop.devzone
15:36:57  <Brot6> 2cc train set - Revision 574:4db2a23ec6f3: Add: 2xShinkansen MU .pcx'es (not ingame yet) (DJNekkid) @ http://dev.openttdcoop.org/projects/2cctrainset/repository/revisions/4db2a23ec6f3
15:44:21  <DJNekkid> yey for me, first commit in a month, ish
15:47:02  <Ammler> DJNekkid: do you already use the "new" nforenum/grfcodec?
15:47:21  <DJNekkid> nope
15:47:25  <DJNekkid> heavent compiled anything yet :)
15:47:26  <Ammler> you should :-)
15:47:32  <DJNekkid> i know :)
15:47:37  <Ammler> there are also binaries
15:47:39  <DJNekkid> just trying to remember where to put it :)
15:49:36  <planetmaker> in your search path
15:53:12  <Ammler> if you like, you can use the rpms from my repo
15:53:34  <Ammler> those are a bit pre-RC1
16:01:15  <Ammler> now, they match RC1 :-)
16:02:30  <Ammler> planetmaker: a new target for the makefile would be nice: spec :-)
16:06:29  <planetmaker> well. Sure. What does it do?
16:08:02  <Ammler> providing a rpm spec file :-)
16:08:35  <Ammler> mainly .devzone/build/grfcodec.spec -> to a spec for releases
16:08:40  <planetmaker> eh? What is non-static about it?
16:09:06  <Ammler> the version
16:12:41  <DJNekkid> wow, all of voyager1's sprites are in the vrong order ... 5-6-7-8-1-2-3-4
16:13:28  <planetmaker> make a new template and be done ;-)
16:14:03  <DJNekkid> haha
16:15:09  <DJNekkid> perhaps if there all were on the same spot in the .pcx
16:15:24  <planetmaker> he, if not, it's not worthwhile, indeed
16:15:31  <DJNekkid> but damn, his artwork is indeed nice!
16:15:35  <planetmaker> I also moved every sprite for se rails.
16:15:39  <planetmaker> yes, it is :-)
16:15:51  <DJNekkid> but a bunch of russian engines inc :)
16:16:16  <DJNekkid> sovjet even
16:17:39  <Doorslammer> Heck, bet his was a better attempt at placement than mine ever was
16:18:32  <DJNekkid> hehe
16:18:41  <DJNekkid> that might actually hold some water...
16:19:01  <Doorslammer> Well, I still haven't worked out what the long blue band was for
16:22:08  <DJNekkid> take a look at the two last revs...
16:22:13  <DJNekkid> very nice gfx
16:22:39  <Brot6> 2cc train set - Revision 575:14a159fa0072: Add: 5 sovjetian/ussr engine pcx-files (DJNekkid) @ http://dev.openttdcoop.org/projects/2cctrainset/repository/revisions/14a159fa0072
16:30:01  <Doorslammer> Oh wow, now those are brilliant
16:41:56  <Doorslammer> I think I see how it's used now
16:42:09  <Doorslammer> Should I update my one?  (again)
16:43:38  <Brot6> 2cctrainset: update from r573 to r575 done (6 errors) - http://bundles.openttdcoop.org/2cctrainset/nightlies/r575
16:44:58  <Brot6> nforenum: update from r463 to r465 done - http://bundles.openttdcoop.org/nforenum/nightlies/r465
16:45:11  <Brot6> Following repos didn't need a nightlies update: 32bpp-extra (r38), airportsplus (r52), basecosts (r20), comic-houses (r71), firs (r1121), fish (r386), grfcodec (r223), heqs (r371), metrotrackset (r43), newgrf_makefile (r124), nml (r670), nutracks (r90), ogfxplus (r41), opengfx (r477), openmsx (r97), opensfx (r97), snowlinemod (r15), swedishrails (r141), transrapidtrackset (r15), worldairlinersset (r659)
16:55:11  <Brot6> Following repos rebuilds successful without any difference to earlier nightlies builds: 2cctrainset (6 errors), 32bpp-extra (1 errors), airportsplus, basecosts, comic-houses (3 errors), firs (142 errors), fish (6 errors), heqs, metrotrackset, newgrf_makefile, nutracks (13 errors), opengfx, snowlinemod, transrapidtrackset, worldairlinersset
17:00:32  <Ammler> it is too silent now :-)
17:01:59  <Rubidium> huh? Why is it too silent?
17:02:32  <Rubidium> no changes == nothing to "worry" about from a dependencies point of view
17:05:04  <Ammler> I would like the Diffsize in brackets like the errors
17:09:22  <Doorslammer> He missed the question
17:09:29  <Doorslammer> Should I update my one?  (again)
17:09:51  <Ammler> DJNekkid: ^
17:16:48  <Doorslammer> Oh well, he can do it then.  Im off to bed
17:16:52  *** Doorslammer has quit IRC
17:27:35  *** Alberth has joined #openttdcoop.devzone
18:34:56  <Brot6> NewGRF Meta Language - Feature #1174: Action 14 support (yexo) @ http://dev.openttdcoop.org/issues/1174#change-3078
19:18:21  <Alberth> planetmaker: something I can do for #1187 (https://dev.openttdcoop.org/issues/1178) and/or #845
19:18:42  <Alberth> I just created a fast hack, but perhaps something better is needed?
19:20:04  <Ammler> Alberth: cool would be a version without any dep check
19:20:56  <Ammler> something switchable
19:21:50  <Ammler> I still have no idea, what the dep check does, what for example grfcodec doesn't
19:24:28  <Ammler> also maybe it could be possible to replace some gnu utils with python scripts, like grep and sed?
19:24:42  <Ammler> so it would again be possible to use the makefile on windows
19:25:26  <Rubidium> vbs!
19:25:48  <Rubidium> not everyone has python installed
19:26:07  <Ammler> hmm, isn't python a requirement for hg?
19:26:26  <Rubidium> is hg a requirement for compiling?
19:26:39  <Ammler> yes,
19:26:51  <Ammler> well, at least for the those which would need it on windows
19:27:12  <Ammler> for the linux guys, we don't need to care
19:28:13  <Ammler> it might also be easier to have linux with wine and mono as a working windows Makefile
19:30:22  <Ammler> Rubidium: or a own c-tool which replaces those?
19:30:36  <andythenorth> Ammler: doesn't dep check try and only parse files that have changed?
19:30:36  <andythenorth> I'm guessing
19:30:57  <Ammler> andythenorth: that does only make sense if you have more than one output file
19:31:04  <Ammler> which only has the basesets
19:31:27  <Ammler> and even there, building evertime everything might be faster then checking deps
19:32:29  <Ammler> but we can keep the dep check as a option
19:37:00  <Alberth> one good question is perhaps, what is  the purpose of a dep check in a newgrf project?
19:37:58  <planetmaker> Alberth, did you test that patch? I feel that it may break...
19:38:11  <Alberth> a bit, not extensively
19:38:16  <planetmaker> But ... yes, I guess dependency check can be made optional this way
19:38:35  <planetmaker> though I'd only guard the lines of depend: itself if ifdef
19:38:44  <planetmaker> then it's save to keep everything else the same
19:38:57  <Alberth> in particular I am not sure that it will notice if you change a .pcx only
19:39:03  <planetmaker> that is: it will be called, but do nothing and return immediately
19:39:11  <planetmaker> it won't then. Of course
19:39:25  <planetmaker> you'll have to use make remake then
19:40:21  <andythenorth> planetmaker: what does the dep check do actually?
19:40:22  <Alberth> you seem to be talking spanish, or perhaps finnish, or at least I don't understand what you say
19:40:27  <planetmaker> I don't understand the need for the first hunk, Alberth
19:40:40  <planetmaker> -include is a conditional include
19:41:01  *** welshdragon has quit IRC
19:41:49  <planetmaker> I guess the question is: how do you want it to work?
19:42:02  <planetmaker> Just compile the grf without any dependencies? Is that right?
19:42:08  <Alberth> one goal is the removal of   'depend' sub-target
19:42:21  <planetmaker> As a result: it won't re-compile, if ANY source file changes
19:42:31  <planetmaker> (if makefile.dep is not present)
19:42:37  <Alberth> the other goal is not to load the .dep file
19:43:27  <planetmaker> That basically will force you to use always 'make remake'. As the target you try to make is already there and depends on nothing but itself
19:43:36  <Alberth> it does, if you don't have sub-targets, make has no way to decide whether it is outdated, so it will get build
19:43:58  <planetmaker> hm.
19:44:14  <planetmaker> there IS actually a sub-target: blub.nfo
19:44:21  <Alberth> with gnu-make you can even make that explicit with the .phony primitive
19:44:21  <planetmaker> which depends upon blub.pnfo
19:44:45  <planetmaker> %.grf -> %.nfo -> %.pnfo
19:44:57  <planetmaker> that's the order by the targets
19:45:57  <planetmaker> but as you very seldom change %.pnfo (the main file), you'll always end up with an up2date newgrf
19:46:19  <planetmaker> as such, you'll then always need to call make remake
19:46:53  <Alberth> well, as I said, it was just a quick hack to see what happens. Now that you are back something more final may be discussed
19:47:50  <Alberth> with my quick & dirty makefiles, I simply list all source files in the makefile, and generate the intermediate targets from that
19:48:21  <planetmaker> I don't know my source files. They're all listed in ... yeah %.pnfo
19:48:34  <planetmaker> and its dependencies
19:49:22  <Alberth> yeah, you'd get a SRC_FILES=.... and    $TARGET : $SRC_FILES  rule
19:49:40  <Alberth> the latter you now do with the dep check generation
19:49:57  <planetmaker> yes
19:50:00  <Alberth> s/the latter/this/
19:51:25  <Alberth> for properly dropping dep checking, a user would have to manually add the source file list in his makefile
19:51:38  <planetmaker> exactly.
19:51:51  <planetmaker> And from my experience that is what the newgrf authors exactly DONT do ;-)
19:51:58  <Alberth> if you don't want that, we should look into a better/faster dep check generation
19:52:00  <planetmaker> that is the whole reason for this automatism
19:52:36  <planetmaker> yes, I concur
19:53:10  <Alberth> hmm, an alternative could be to query the file system for all source files
19:53:44  <Alberth> ie something like SRC_FILES=`ls *.pnfo`
19:53:49  <planetmaker> hm. *.pnfo *.tnfo *.png *.pcx and no inter-dependency?
19:53:59  <planetmaker> that will be WAY quicker
19:54:20  <planetmaker> But... it will also in my usual repos pick up a lot of garbage ;-)
19:54:27  <Alberth> dependencies is   %.pnfo: *.png *.pcx
19:54:37  <Alberth> or so
19:54:38  <planetmaker> though... maybe it's a good solution
19:54:54  <planetmaker> just %.grf: %.pnfo %.pcx %.tnfo
19:55:05  <planetmaker> make the grf directly depend upon it.
19:55:09  <planetmaker> hm. No.
19:55:15  <planetmaker> The nfo
19:55:18  <planetmaker> %.nfo: ...
19:56:49  <Alberth> the disadvantage is a full recompile for any change in a .pcx file, but that does not seem to be a problem
19:57:29  <planetmaker> Depends... we need to pick the dirs with grf-pcx files and those which just contain 'sources'
19:57:32  <planetmaker> but that's mostly done
20:02:42  <Alberth> $(wildcard *.c)   is listed as the way to get all .c files in a gnu-make makefile
20:02:57  <planetmaker> that can be changed.
20:03:08  <planetmaker> I also search all paths. So that's not the issue, yes
20:04:48  <Alberth> is there something I can do (in the weekend) to help you with this?
20:05:40  <planetmaker> I guess, the proposal to just check for the presence of the source files in the sprite subdirs, is as good as we can do, if we don't want to scan the files
20:06:33  <planetmaker> It would need checking how it performs...
20:06:41  <Alberth> I don't know why scanning takes so much time.
20:06:43  <planetmaker> especially on windows
20:06:56  <planetmaker> grep etc is very slow on windows obviously
20:07:41  <Alberth> yeah, so if you'd use eg a python program, the whole issue may be solved already
20:07:54  <Alberth> but is it worth the trouble?
20:08:30  <Alberth> I suspect forking so many processes is what kills the performance
20:08:34  <planetmaker> might be. Any hg repo requires python. So a python dep check would be fine, too
20:09:18  <planetmaker> basically it's what it is now (bash) as it worked for me and others until windows started to complain (again) about hte performance.
20:09:32  <planetmaker> But the latter hadn't really been tested on windows since my last rewrite of it.
20:09:44  <planetmaker> I wouldn't mind a python script at all.
20:11:26  <planetmaker> if you look at openmsx you might see that I work there with small python scripts for processing, too
20:11:31  <Alberth> hmm, would restricting the set of source files to those tracked by hg be useful?
20:11:43  <planetmaker> not really
20:12:08  <planetmaker> when I add a new feature, i might add a new file. But not yet add it to hg while testing / writing
20:12:11  <Alberth> I use python for everything :)
20:12:15  <planetmaker> hehe
20:13:20  <Alberth> eg looking for a bad pattern in the C++ source code  so I can make a codechange that touches lots of files :p
20:14:11  <planetmaker> by my standards you're a python guru ;-)
20:14:29  <planetmaker> you're on windows, right?
20:14:41  <andythenorth> python keeps me happy :P
20:15:03  <Alberth> the only windows I have are those managed by my X11 server :)
20:15:10  <Alberth> and in my house :)
20:15:15  <planetmaker> oh. he.
20:15:49  <planetmaker> well. As said: I'm happy to replace the dep check by depcheck.py or makedep.py
20:16:33  <Alberth> ok, you give the script one or more file names, it scans them, and produces dependencies for those files?
20:16:50  <Alberth> should it also scan newly found files?
20:16:57  <planetmaker> If you want to give it a shot, I'll happily then change that, if it makes foobar happy. Myself I won't have too much time at hand this weekend, but... commiting that to the repos is then not too much work
20:17:24  <planetmaker> Alberth, best would be a recursive scan. Starting with the file given. And then its dependencies as found
20:17:56  <Alberth> perhaps I should checkout the project that foobar works on, and use that as test case :)
20:17:57  <planetmaker> ideally like makedep filename.pnfo > makefile.dep
20:17:59  <planetmaker> or so
20:18:08  <planetmaker> firs is certainly a good test case
20:18:25  <planetmaker> it's the worst, next to 2cctrainset
20:18:30  <planetmaker> 'worst'
20:18:40  <Ammler> planetmaker: you should answer the "main question", what is the dep check for?
20:19:18  <planetmaker> you mean: unconditionally re-build
20:19:45  <Ammler> why does your Makefile need it?
20:20:05  <Alberth> in that case, why do you need a makefile?
20:20:15  <planetmaker> :-)
20:20:18  <Ammler> yes, why?
20:20:30  <planetmaker> Ammler, in order to not build, if it doesn't need building
20:20:48  <planetmaker> which makes especially sense for opengfx
20:21:01  <planetmaker> as I have 6 newgrfs otherwise always built fully, if I work on only one
20:21:32  <Ammler> yeah, opengfx is a special case, let us ignore it for this question
20:23:06  <Ammler> what is the advantage of the Makefile for ser?
20:23:08  <Alberth> Ammler: throwing away a makefile is easy, but then you need some script to build the grf, and you don't want to write that by hand
20:23:57  <Ammler> well, IMO, it woudn't be more complicated then writing the Makefile, is it?
20:24:27  <planetmaker> Ammler, make is made to define the exact order of processing steps in order to build something
20:24:34  <planetmaker> and to only do those which need doing
20:24:49  <Ammler> which is always everything or nothing
20:24:51  <planetmaker> so...?
20:24:54  <Ammler> in our case
20:24:59  <planetmaker> Ammler, it's not
20:25:03  <Ammler> when?
20:25:20  <planetmaker> documentation + grf are separate. md5 is separate. bundles are separate
20:25:33  <planetmaker> so this is about 5 different targets at least
20:25:37  <andythenorth> building FIRS cargo list with awk is separate?
20:25:42  <Ammler> yep, but only one which needs time
20:25:57  <planetmaker> andythenorth, no. It's a dependency. But it's a separate target.
20:26:09  <planetmaker> so... if it's not touched, then it's not re-build, yes
20:26:19  <planetmaker> (or its dependencies not touched)
20:26:28  <Ammler> why shall I run make, if I didn't touch it?
20:26:29  <planetmaker> so... actually: yes :-P
20:26:51  <planetmaker> Ammler, because you're a lazy brag, have a bad memory and just call make to get the newest version
20:26:54  <planetmaker> ;-)
20:27:09  <Ammler> yep, in that case, I wouldn't care to build it again
20:27:16  <Ammler> without dep check this is done in 3 secs
20:27:27  <Ammler> the dep check needs 10 secs
20:27:32  <planetmaker> now my question: why shall I write it as a script?
20:27:44  <Ammler> to skip the dep check
20:28:02  <Ammler> to have a syntax which someone else could help maintaining
20:28:04  <planetmaker> for that means I don't need a script. I CAN also write a makefile which always re-builds
20:28:17  <planetmaker> Makefile syntax is quite common knowledge
20:28:34  <planetmaker> and not that different from bash actually
20:28:34  <Ammler> well, it seems nobody is able to disable dep check
20:29:06  <planetmaker> As Alberth lined out: it's possible. And yes, one could always re-build
20:29:49  <Ammler> actually, disabling the dep check might make it able to build on windows again?
20:30:37  <Ammler> or is grep somewhere else used?
20:31:06  <Ammler> or is there something else than grep, which fails on windows?
20:32:52  <Alberth> process forking :)
20:33:07  <Alberth> it takes too long
20:35:04  <Ammler> if nml could handle includes and substitutes, you could through away the Makefile
20:35:32  <planetmaker> no
20:35:42  <planetmaker> or what would create the bundles?
20:35:50  <planetmaker> and install the thing?
20:36:18  <planetmaker> I don't quite get why you want to throw away make just because the _current_ dep check is slow
20:36:35  <Ammler> not just that, it is not working on windows
20:36:54  <planetmaker> make works on windoze
20:37:05  <Ammler> not this Makefile
20:37:22  <planetmaker> so a bash script would work better?
20:37:48  <Ammler> not if it would use the same tools
20:37:54  <planetmaker> I've not seen anywhere a report that it fails with <whatever> under windows
20:37:56  <Ammler> so rather something pythonish
20:38:19  <Ammler> the thing with grep was never reported?
20:38:48  <Ammler> really new to you?
20:39:04  <Ammler> then I am sorry :-(
20:39:31  <Ammler> the was Makefile for example works on windows
20:39:51  <planetmaker> then it's probably grep -o which doesn't work
20:39:59  <planetmaker> which is a matter of the grep version in use
20:40:16  <Ammler> yep, which needs to be very old because of mingw
20:40:33  <Ammler> dunno, if that is really the only thing
20:41:42  <planetmaker> http://dev.openttdcoop.org/projects/newgrf-makefile/issues <-- in any case: I have no bug report about that
20:41:43  <Ammler> I know, you invested a lot time for the dep check, does it really make the Makefile useless, if you through that away?
20:41:58  <planetmaker> no, it doesn't
20:42:12  <planetmaker> it just needs to be replaced by *something*
20:42:20  <Ammler> why?
20:42:22  <planetmaker> and that *something* can be:
20:42:32  <Ammler> failed building = failed dep check
20:42:42  <planetmaker> - what Alberth and I talked about: a rule to depend on just all source files in certain dirs
20:42:54  <planetmaker> - a python script writing the rules
20:43:18  <Ammler> I guess building errors might also be more verbose then dep errors
20:44:45  <planetmaker> not really
20:45:04  <planetmaker> and last alternative is: rebuild always rule
20:45:20  <Ammler> yes, which most here would prefer
20:45:34  <Ammler> well, you could compare last build rev with current rev
20:46:01  <Ammler> but grepping all the files is really just wasting time
20:46:46  <Ammler> it is indeed a very cool thing, but simply no advantage in _our_ case
20:48:38  <Ammler> people use it to respect your work :-)
20:49:03  <planetmaker> ...
20:49:13  <Alberth> and it is too complicated to make something yourself
20:49:56  <Ammler> yes, but nobody ever liked the dep check
20:50:12  <andythenorth> didn't it bring some advantages?
20:50:37  <andythenorth> I forget what, but I thought I was glad when it was added...
20:51:02  <planetmaker> if you change the readme and compile a new version, there's no point to re-compile the grf
20:51:12  <Ammler> if you have more than one output file, which isn't the case here
20:51:17  <planetmaker> if you modify a source image which is not included. There's no point to re-compile the grf
20:51:42  <Ammler> it does
20:51:44  <planetmaker> if you change a testing file not yet included, no need to re-compile
20:51:53  <Ammler> since you force rebuilding because of rev change
20:52:16  <planetmaker> Ammler, the dep check is mostly sensible when _developing_ newgrfs.
20:52:28  <planetmaker> E.g. when you're working with the repo. For me. for andy. for djn
20:52:34  <planetmaker> when we actually work on the newgrf
20:52:42  <planetmaker> when we constantly change one thing or another
20:52:50  <planetmaker> all of those three.
20:52:54  <Ammler> then you know, if you need make or not
20:53:07  <andythenorth> it's useful e.g. when you're not sure if something changed in game
20:53:12  <planetmaker> ^
20:53:31  <Ammler> andythenorth: please explain
20:53:41  <planetmaker> just running 'make install' and then ingame a newgrf reset and you know you're testing the current version
20:53:49  <andythenorth> exactly
20:54:20  <Ammler> and what is the difference, if you would just rebuild
20:54:32  <andythenorth> faster
20:54:40  <Ammler> what is faster?
20:54:42  <andythenorth> if there's nothing to build, it's pretty much instant
20:55:07  <Ammler> the dep check needs around double the time as a simple rebuild
20:55:44  <andythenorth> FIRS make install is 2.8s here if there is nothing changed
20:55:54  <planetmaker> Ammler, for the logic train: yes. But not for FIRS + 2cctrainset
20:57:21  <andythenorth> I'm so used to dep check, I'm not sure what it would be like without it
20:58:15  <Alberth> I see no problem in providing 2 options with make, either assume all files are part of the source, or doing a dep check
20:58:46  <Alberth> but imho we should first see what happens if you have a python script for the dep check calculation
20:59:06  <andythenorth> in the case of my *old* FIRS repo, 'all files part of source' would have been a horrible assumption :)
20:59:16  <Alberth> perhaps it is so fast that the other option is useless
20:59:16  <andythenorth> but since I destroyed my repo with strip, that's no longer the case :P
20:59:48  <Ammler> well, you need to work with it, IMO, it is useless without multiple output files
21:00:08  <Ammler> and sorry, I don't count a readme as a output file :-P
21:01:08  <frosch123> a makefile is also some kind of script, whjich might be easier than a shell script
21:01:17  <Ammler> or complicater
21:02:23  <Alberth> compared to manually maintaining a script, easier by a long shot
21:02:36  <andythenorth> list of dependencies is a file?
21:03:01  <Alberth> look at the makefile.dep
21:03:07  <planetmaker> andythenorth, yes
21:03:33  <andythenorth> does it just compare timestamps?  Or more?
21:03:53  <Alberth> timestamps and existence of the file :)
21:04:28  <Ammler> a full make needs 22 secs, where depend is 16 secs
21:04:39  <Ammler> of firs
21:05:00  <Alberth> Ammler: don't use the current implementation as base for decisions
21:05:18  <Alberth> we all know it is not fast enough
21:05:27  <Ammler> it would only make sense, if it is at least less then building itslef
21:05:46  <Ammler> it is just not logical for me :-)
21:05:57  <planetmaker> Ammler, yes. And that we all agreed upon.
21:06:15  <planetmaker> But that doesn't make your concluse "It's useless" valid.
21:07:06  <andythenorth> seems a long time to just compare timestamps
21:07:25  <andythenorth> hmm
21:07:32  <Ammler> andythenorth: in cases where you have to build something
21:07:41  <andythenorth> on the mac there would be various crazy options to update dep file when saving
21:07:44  <Ammler> which is in 99% the case
21:09:11  <Alberth> anyways, good night for now
21:09:22  <andythenorth> bye Alberth
21:09:28  <Ammler> planetmaker: does the check for readme change make it valied?
21:09:42  <planetmaker> night Alberth
21:10:01  <planetmaker> please re-phrase, Ammler
21:10:16  <Alberth> Ammler: you need to replace the makefile by something else automagic
21:10:20  *** Alberth has left #openttdcoop.devzone
21:10:26  <Ammler> that is your "advantage" of the dep check, so you don't need to rebuild if you have changed the readme
21:11:10  <Ammler> well, we could also talk about the dep check in openttd, where you have a lot output files
21:11:26  <Ammler> but on 99% of my makes it does rebuild everything
21:11:55  <planetmaker> I hardly rebuild everything each time
21:12:18  <Ammler> you need around 10 commits to reach that :-)
21:12:29  <Ammler> hmm, fewer
21:12:44  <Ammler> but there it makes sense, if you patch something
21:12:57  <planetmaker> it's made for the case where you _work_ on the newgrf
21:12:58  <andythenorth> planetmaker: this is my impression, not sure if it's true....if I only change a pcx, it doesn't run cpp to build a new firs.nfo?
21:13:09  <andythenorth> seems to be the case when I use it
21:13:12  <planetmaker> andythenorth, that's true
21:13:18  <planetmaker> why should it?
21:13:19  <Ammler> planetmaker: but you rebuild evertime the whole grf :-)
21:13:30  <planetmaker> the nfo doesn't change. Just the image. So only grfcodec is re-run
21:13:33  <andythenorth> but not all the (many many) cpp files
21:13:38  <planetmaker> or at least that makes sense
21:13:57  <andythenorth> Ammler: I am somewhat influenced by FIRS which has *lots* of templates to build / process
21:14:38  <Ammler> andythenorth: you would see the difference, if it would be possible to disable depend check
21:14:53  <andythenorth> perhaps
21:15:02  * andythenorth is on the fence
21:15:11  <Ammler> and then it would be interesting, if you would miss it
21:15:13  <Ammler> and why
21:15:37  * andythenorth should go to sleep
21:15:50  <Ammler> planetmaker: in nfo/grf case, you still have the advantage
21:15:57  <andythenorth> speed of FIRS make file not currently biggest / most interesting question in my life :)
21:16:00  <Ammler> since those are 2 different targets
21:17:14  <andythenorth> full make clean of FIRS tip is 11.6s for me
21:17:22  <Ammler> I also don't think, it matters, if you would use a 10 sprites grf or a 10k sprites
21:17:40  * andythenorth bed
21:17:43  <andythenorth> good night
21:17:46  <Ammler> nightly
21:20:53  <Ammler> planetmaker: people which like the depend check, could simply run "make depend && make"
21:34:10  *** welshdragon has joined #openttdcoop.devzone
21:36:37  *** frosch123 has quit IRC
21:40:23  *** Ammler has quit IRC
21:40:23  *** Hirundo has quit IRC
21:40:23  *** V453000 has quit IRC
21:41:11  *** Ammler has joined #openttdcoop.devzone
21:41:11  *** Hirundo has joined #openttdcoop.devzone
21:41:11  *** V453000 has joined #openttdcoop.devzone
22:15:18  *** ODM has quit IRC
22:33:12  *** thgergo has quit IRC

Powered by YARRSTE version: svn-trunk