--- Log opened Wed Oct 10 00:00:07 2007
--- Day changed Wed Oct 10 2007
00:00 < dooglus> non-repeatable, apparently random
00:00 < genete> yes
00:00 < dooglus> your nested-canvases bug
00:00 < genete> some straneg link or similar things
00:00 < dooglus> I think it can be fixed by changing:
00:00 < dooglus> 		if(area.area()<=0.000001 || desc.get_w()==0 || desc.get_h()==0)
00:00 < dooglus> to
00:00 < dooglus> 		if(area.area()<=0.000000 || desc.get_w()==0 || desc.get_h()==0)
00:00 < dooglus> in layer_pastecanvas.cpp
00:00 < genete> ah man
00:01 < genete> I saw that line before
00:01 < dooglus> you did?
00:01 < genete> I don't like usage of numeric constants in the code
00:01 < dooglus> no
00:01 < genete> I have readed some changes you made
00:01 < genete> I prefer to make a define and modify it at the .h file
00:02 < dooglus> 'enum' is the preferred way in C++ I think
00:02 < dooglus> especially when it's like 0=NEAREST NEIGHBOUR, 1=LINEAR, 2=COSINE, etc.
00:02 < genete> it doesn't matter but place 0.00001 is a bad behaviour
00:02 < genete> why not 0.00000001
00:02 < dooglus> certainly don't use "2" in the code though, and expect the reader to remember that it means cosine
00:03 < dooglus> I think it's to do with the inaccuracy of floating point numbers
00:03 < genete> have you converted a list parameter to Reference
00:03 < dooglus> I think so
00:03 < genete> the Link is an integer
00:04 < genete> so it is not so weird
00:04 < genete> to have integers inside a converted sub-parameter to substitute the lists
00:04 < dooglus> I don't see an integer
00:05 < dooglus> I see it as a list
00:05 < genete> Conver the Feather parameter to Reference
00:05 < dooglus> ok
00:05 < dooglus> I see
00:05 < dooglus> then set the value to 8
00:05 < dooglus> ...
00:06 < genete> maybe you can make an automatic export and link when want to convert to random type...
00:06 < dooglus> (only 0 through 4 are valid)
00:06 < dooglus> I saw you saying about this yesterday and didn't understand
00:07 < genete> When you need a list in the sub parameter just export a listed parameter and link to it an integer
00:07 < dooglus> this isn't changing types - 'type of feather' was always an int
00:07 < genete> I know
00:08 < dooglus> ah - I misunderstood.
00:08 < genete> but this would allow the user have some meaning to that integer
00:08 < dooglus> you are talking about the random convert...
00:08 < genete> yes
00:08 < dooglus> I forgot about random converts
00:08 < genete> the idea comes from it
00:08 < genete> Don't worry
00:08 < dooglus> ok, I'm understanding now
00:08 < genete> put in a corner of your brain
00:08 < dooglus> does linking work?
00:09 < genete> I've not tried
00:09 < dooglus> it doesn't
00:09 < genete> beacuse I have not lists subparameters in the conversion types
00:09 < dooglus> because when you export a 'type of feather' it becomes an 'int' in the child dialog
00:09 < dooglus> and when you link to it, it stays as just an int
00:10 < dooglus> it forgets anything about feathers
00:10 < genete> wait
00:11 < dooglus> for how long?
00:11 < genete> hehe
00:12 < genete> you can still using the original parameter and the exported change its value
00:12 < dooglus> yes
00:12 < genete> I've converted and then exported the main parameter
00:12 < dooglus> did you notice that this is mentioned in the TODO file?
00:12 < genete> than the exported follows the main parameter
00:13 < dooglus> in the synfig-core TODO:
00:13 < genete> now link a subparameter to the exported and ...
00:13 < genete> what?
00:13 < dooglus> "* Make layer and value-node parameters handled in the same way. (currently different)"
00:13 < genete> ah I never understood the meaning of that
00:14 < dooglus> I'm pretty sure that's what we're talking about.  layers have parameters, and value-nodes have parameters - they're what we've been calling 'sub-parameters'
00:15 < genete> I see
00:16 < genete> dooglus: comming back to single threaded core
00:16 < genete> If I apply the patch over new svn would it work?
00:17 < genete> I want to have the latest svn and also test the patch to see if random crash dissapear
00:20 < dooglus> that's what you did I think
00:22 < genete> ok, it's my bed time.
00:22 < dooglus> goodnight
00:22  * genete is leaving... Good night all
00:22 < dooglus> how does the single-threaded version seem to you?
00:22 < dooglus> first impressions?
00:22 < genete> with normal files looks the same   I think
00:22 < dooglus> ok, well that's good
00:23 < genete> I'll teel you if I find some thing weird
00:23 < genete> night all!
00:23 < dooglus> thanks
00:23 < dooglus> goodnight
00:23 -!- genete [n=Genete@84.122.44.48.dyn.user.ono.com] has quit ["Abandonando"]
00:24 < dooglus> are people happy enough with -rc3 to make it the final release?
00:25 < dooglus> I'd like to get it out the way so I can carry on committing new stuff to svn
00:34 < KiBi> dooglus: FWIW, I'm fine. :)
00:34 < KiBi> Waiting for pabs3 for the following uploads. :)
00:34 < dooglus> ok
01:05 < ulrik> 'night!
01:06 -!- ulrik [n=ulrikbod@81-231-118-204-no53.tbcn.telia.com] has quit ["bye! http://www.musikboden.se"]
01:21 -!- pabs3 [i=daemon@60.242.186.48] has joined #synfig
01:37 < pabs3> KiBi: I backed out of the arch all thing, because of /usr/lib/pkgconfig/ETL.pc - we should change that in svn.voria.com to /usr/share/pkgconfig
01:37 < KiBi> pabs3: Yep, I was waiting for you for that.
01:38 < KiBi> Still built on only 5 archs. :/
01:38 < pabs3> looks like other packages are in front: http://buildd.debian.org/~jeroen/status/package.php?p=etl
01:40 < pabs3> hmm, one advantage of being arch any is that the tests are run on all arches. if we could get the result of the fixed test ignored, then this would be useful
01:42 < dooglus> pabs3: does synfig build on arm now?  I saw you mention it in the debian BTS
01:42 < pabs3> seems so: http://buildd.debian.org/~jeroen/status/package.php?p=synfig
01:43 < pabs3> studio did too, so that means it works there too :) http://buildd.debian.org/~jeroen/status/package.php?p=synfigstudio
01:43 < dooglus> triffic :)
01:43 < KiBi> pabs3: Oh, yet, that (tests).
01:43 < KiBi> *yes
01:44 < dooglus> pabs3: did you see the new comments on http://bugs.kde.org/show_bug.cgi?id=137396 ?
01:44 < KiBi> do we have hppa troubles, now that I have an account on a cluster? :D
01:45 < KiBi> Oh, Lukas Fittl there.
01:45 -!- ZanQdo [n=Daniel@201.201.2.22] has quit ["Adios"]
01:56 < dooglus> would it be useful if the 'fixed' ETL test passed?
01:57 < dooglus> I mean, does the fact that it fails actually affect anything on the debian side?
01:57 < KiBi> dooglus: That's visible in the build logs, but that doesn't make the build fail.
01:57 < KiBi> Which means: that'd be useful if someone checks the log. :)
01:58 < dooglus> KiBi: the code that's failing isn't used anywhere in synfig, so we can safely 'relax' the tests
01:58 < dooglus> (in the same way that the 'rm' command relaxes a file)
01:59 < KiBi> :-)
02:00 < dooglus> what's a good way of timing a function?
02:00 < dooglus> I'm running it millions of times and counting elapsed time, but it varies every time I run it
02:01 < KiBi> From x to y?
02:01 < KiBi> I mean: which values of x and y?
02:02 < dooglus> there's a 'fixme' comment in the ETL code that rounds a float to an int
02:02 < dooglus> I found a library called 'vtk' which claims to have a fast function to do that
02:02 < dooglus> but in timing tests, sometimes it's faster and sometimes it's slower than the existing method, on the same data
02:02 < KiBi> Increase the number of millions, then?
02:03 < dooglus> ok...
02:03 < dooglus> I imagine that will just increase the differences though
02:03 < dooglus> I'm doing 200 million already
02:04 < KiBi> What time range does it take?
02:04 < dooglus> 10 seconds
02:04  * KiBi launching another openscenegraph build, BBS.
02:04 < dooglus> 9 to 11
02:05 < KiBi> Not very significant to me. You've got the startup overhead, the scheduling stuff, etc., which may impact your test.
02:06 < KiBi> Letting run for something like a couple of minutes would ensure that such effects no longer happen.
02:20  * pabs3 building synfig
02:30 < dooglus> I'm calling time() twice inside the code, so startup overhead shouldn't come into it
02:30 < dooglus> I'll try using the 'time' command instead, so I can ignore 'real' time
02:33 < dooglus> on 2 successive runs, the cpu time for the same code changed from 1m11s to 1m21s
02:34 < dooglus> that's like a 15% difference
02:34 < dooglus> maybe 'speedstep' is having an effect - as the CPU gets hotter, it steps down the speed?
02:36 < KiBi> maybe; performance measurement isn't trivial /o\
02:37 < dooglus> heh
02:37 < dooglus> I tried with gcc -O4
02:37 < dooglus> at that point the compiler realised I was only doing a silly performance test and ran all the tests in 0.000 seconds :)
02:38 < KiBi> ahahaha
02:45  * pabs3 building synfigstudio
02:49 -!- pxegeek [n=chatzill@c-71-59-140-184.hsd1.or.comcast.net] has joined #synfig
02:53 -!- crazy_bus [n=philip@CPE-124-184-230-3.nsw.bigpond.net.au] has joined #synfig
03:08 < pabs3> got a crash with rc3 jumping to different times in eye.sifz
03:10 < pabs3> hhmmm, lots of synfig(18703): warning: inputdevice.cpp:94 DEBUGPOINT
03:11  * pabs3 would have thought DEBUGPOINTS would only show up when doing a debug build
03:13 < pabs3> anyways, gonna build, tag & upload synfig now 
03:17 < CIA-28> synfig: pabs * r870 /synfig-studio/trunk/synfigstudio.desktop.in: Remove obsolete Encoding field from .desktop file
03:19  * dooglus realises that the reason that fixed-point was doing so poorly compared to floating point in the ETL speed test is that the compiler was short-cutting the floating point tests...
03:19 < dooglus> so the fixed-point stuff is faster after all - about 2000% faster (?)
03:23 < pabs3> :)
03:23 < pabs3> how do you mean short-cutting?
03:24 < dooglus> I mean the optimizer realises that we're doing a million additions, but not using the result, so it doesn't bother generating code to do the additions
03:24 < pabs3> ahh, ok
03:24 < dooglus> or it realises the at some point we're multiplying by zero, and knows that it needn't bother generating code to calculate what to multiply by zero
03:25 < pabs3> are you thinking -O0 or change the code?
03:25 < dooglus> I'm changing the code
03:25 < dooglus> the results are useful
03:25  * pxegeek clicking in eye timeline....
03:25 < dooglus> there's one bit where it multiplies by 0.25
03:25 < dooglus> in the integer test, that's 0, so it's quick...
03:26 < pabs3> pxegeek: I couldn't reproduce it, prob another race thingi. also got a hang on shutdown I couldn't reproduce either
03:26 < dooglus> pxegeek: I've got the single-threaded patch ready if you want to try it on windows
03:26 < pabs3> yay for uploading -dbg packages
03:26 < pxegeek> I saw the patch, but like genete, don't know if I can apply it.
03:26 < dooglus> pxegeek: http://dooglus.rincevent.net/synfig/single-threaded-patch.txt
03:26 < dooglus> but I made .tar.gz files if you would rather
03:27 < pxegeek> I saw in the logs
03:27 < dooglus> http://dooglus.rincevent.net/synfig/synfig-0.61.07-single-threaded.tar.gz
03:27 < dooglus> http://dooglus.rincevent.net/synfig/synfigstudio-0.61.07-single-threaded.tar.gz
03:27 < pxegeek> That would suck b/w but be easier
03:27 < dooglus> only a couple of MB
03:27 < pabs3> KiBi: synfig all uploaded
03:28 < KiBi> pabs3: ok, thanks
03:28 < dooglus> pxegeek: no change to ETL
03:28 < pabs3> I'll wait till synfig is in the archive before doing synfigstudio
03:28 < dooglus> pxegeek: the patch assumes you have a directory structure like in svn, with 'trunk' directories
03:28 < pxegeek> Hey - stop typing things - I'm trying to click here!
03:29 < pxegeek> I think it also references git
03:29 < dooglus> sorry.  thought this was IRC.  wrong window?
03:29 < pxegeek> ;)
03:29 < dooglus> pxegeek: the '-p1' argument to patch tells it to ignore the 1st dir
03:29 < pxegeek> d/l now
03:29 < dooglus> pxegeek: it worked ok for genete anyway, after those 3 'ln' commands.
03:30 < dooglus> doubt 'ln -s' works too well on windows though
03:30  * pxegeek going to read to my son
03:30 < pxegeek> ln don't work so good for me :(
03:30 < dooglus> pxegeek: you've got an hour or two while it rebuilds :)
03:40 < pxegeek> unfortunately didn't get a chance to hit the go button before I had to go though...
03:40 < dooglus> pxegeek: I've left in a 'print' command whenever it starts rendering
03:41 < dooglus> letting you know you're running the single-threaded version
03:41 < dooglus> ('cos it's hard to tell otherwise)
03:41 < pxegeek> I just killed synfig by trying to render Mandelbrot at 2000x3000
03:41 < pxegeek> :)
03:41 < dooglus> if you comment out the line that says:
03:41 < dooglus> #define SINGLE_THREADED
03:41 < dooglus> in studio/src/gtkmm/asyncrenderer.h
03:41 < dooglus> then it'll go back to normal
03:43 < dooglus> this is funny:
03:43 < dooglus> calculating float...float time: 5.811273sec
03:43 < dooglus> calculating fixed...fixed time: 0.276704sec
03:43 < dooglus> Fixed point wins by 5.534569 seconds! (2000.176723% faster)
03:43 < dooglus> darco reckoned it would be 2000% faster
03:43 < dooglus> but it's 2000.1% faster!
03:44 < KiBi> :D
03:45 < pabs3> heh
03:45 < pxegeek> Maybe he did the same test!
03:46  * pxegeek building singlethread core
03:50 < pxegeek> pabs - have you heard anything from ChrisVP recently?
03:50 < pabs3> nope
03:50 < pxegeek> I'm wondering if he ever finished the design
03:55 < pabs3> his synfig sites seem unchanged
03:56 < pabs3> I'd like us to use mediawiki (initially, and then ikiwiki) rather than drupal for the site anyway
03:57 < pxegeek> Can we use mediawiki stylesheets on ikiwiki?
03:57 < pxegeek> s/stylesheets/skins
03:58 < pabs3> not sure, probably unlikely
03:58  * pxegeek just crashed eye, but I had to click an in sane amount of times on the timeline to do it.
04:02 < pabs3> one migration path for mediawiki -> ikiwiki would be to write an ikiwiki plugin using libmediawiki-perl that would process .mediawiki files. this would be good for wiki users too, cause they wouldn't have to learn markdown
04:02 < CIA-28> synfig: dooglus * r871 /ETL/trunk/ETL/_fixed.h: Conditionally compile in the '+0.5' code to make fixed point numbers more accurate. Currently disabled.
04:02 < CIA-28> synfig: dooglus * r872 /ETL/trunk/test/fixed.cpp: 
04:02 < CIA-28> synfig: Various changes: Improve the formatting of the output. Stop the compiler
04:02 < CIA-28> synfig: optimizing out the calculations. Allow greater error when multiplying by PI if
04:02 < CIA-28> synfig: _fixed.h isn't rounding to the nearest integer. Define variable before using
04:02 < CIA-28> synfig: it. Tidy the source code layout. Prevent a core dump by not allowing numbers
04:02 < CIA-28> synfig: to wrap around to zero.
04:08 -!- Zelig [n=Zelig@ip68-108-123-29.lv.lv.cox.net] has joined #synfig
04:13  * pxegeek applauds Zelig for his animation!
04:15 < Zelig> pxegeek: Thanks. It's colse to being finished. But I'm about to lose my mind over Cinelerra.
04:16 < Zelig> It's crashing while trying to open the file I previously created.
04:16 < pabs3> egads
04:16 < pxegeek> Eeeek!
04:17 < Zelig> I know we all like to support open souce programs, and I feel great to be a part of a community like this that supports and encourages open-source pregrams..
04:17 < CIA-28> synfig: dooglus * r873 /ETL/trunk/test/hermite.cpp: #include <ETL/fixed> so ETL/_bezier.h doesn't have to.
04:17 < CIA-28> synfig: dooglus * r874 /ETL/trunk/ETL/_bezier.h: Don't include <ETL/fixed> because it's not used.
04:17 < Zelig> But a program that crashes while opening its OWN file format?!
04:17  * pxegeek thinks his compile will go a lot quicker if I rename the folder correctly!
04:17 < Zelig> Ye flippin gods.
04:19 < Zelig> I'm trying to install Kdenlive right now.
04:19 < pxegeek> Ah - much better
04:19 < Zelig> </rant>
04:21 < pxegeek> now what?
04:22 < pxegeek> Automake error :(
04:24 < pxegeek>   /bin/sh /home/Chris/synfig/build/synfig-core/libltdl/missing --run automake-1.7 --foreign  Makefile
04:24 < pxegeek> configure.ac:53: version mismatch.  This is Automake 1.7.1,
04:24 < pxegeek> configure.ac:53: but the definition used by this AM_INIT_AUTOMAKE
04:24 < pxegeek> configure.ac:53: comes from Automake 1.7.3.  You should recreate
04:24 < pxegeek> configure.ac:53: aclocal.m4 with aclocal and run automake again.
04:24 < pxegeek> I've seen version mismatches before but they're usually not fatal
04:25 < pabs3> another ./bootstrap run should fix it
04:26  * pxegeek starts hyperventilating
04:31 < pxegeek> I reran the same script file and it seems happier now.  
04:32  * pxegeek can't see why I don't get the same result the second time, but as long as it's working <shrugs>
04:32 < pxegeek> biab
04:36  * dooglus is going to make a 0.61.07 final release now
04:36 < dooglus> enough delaying, right?
04:38 < pxegeek> No less broken than 0.61.06
04:38 < pxegeek> All in favor?  Say aye!
04:38 < Zelig> Arrr!
04:38 < Zelig> Sorry... Aye!
04:39 < dooglus> always best to make a release after a quick succession of updates to the core template library, I say
04:39 < pabs3> heh
04:40 -!- crazy_bus [n=philip@CPE-124-184-230-3.nsw.bigpond.net.au] has quit [Remote closed the connection]
04:42 < pxegeek> Does this mean that the windows build needs to be recompiled again?  Or can we rename RC2 (he says hopefully)
04:42 < dooglus> RC3 you mean?
04:42 < pxegeek> No......  
04:42 < pxegeek> I never got to build RC3.  You said we wouldn't get there ;)
04:43 < dooglus> right
04:43 < dooglus> the NEWS files are updated with the date and the svn release number
04:43 < dooglus> so the .exe installers should be rebuilt to get that change
04:44 < pxegeek> OK - I'll get the tarballs when they're posted - if Atrus doesn't beat me to it.  
04:44 < dooglus> they're posted
04:44  * pxegeek sends Atrus a hug
04:44 < dooglus> well, the 1st 2
04:44  * pxegeek is compiling single threaded right now - won't kick off another compile tonight
04:45 < pxegeek> Early start tomorrow, and a long day
04:45 < pxegeek> 7am - 6pm :(
04:47 < dooglus> ok, it's all up on sourceforge now
04:47 -!- GustoEater [n=doug@cpe-76-169-178-220.socal.res.rr.com] has joined #synfig
04:47 < CIA-28> synfig: dooglus * r875 / (3 files in 3 dirs): Updated svn revision number and date in NEWS files.
04:47 < CIA-28> synfig: dooglus * r876 /ETL/tags/ETL_0_04_10: Release ETL_0_04_10
04:47 < CIA-28> synfig: dooglus * r877 /synfig-core/tags/synfig_0_61_07: Release synfig_0_61_07
04:47 < CIA-28> synfig: dooglus * r878 /synfig-studio/tags/synfigstudio_0_61_07: Release synfigstudio_0_61_07
05:04 < dooglus> updated http://www.synfig.com/
05:07 < dooglus> updated sourceforge news, too
05:07 < dooglus> does someone want to write something for LWN?
05:07 < dooglus> I'll update freshmeat tomorrow.
05:07 < dooglus> goodnight all :)
05:08  * pabs3 updates gnomefiles
05:08 < dooglus> oh, I forgot to add a message in the console when studio is starting as pxegeek suggested.
05:08 < dooglus> there's always 0.61.08, eh?
05:12 < dooglus> freshmeat: "Your new release (ID 263403) has been stored for verification by the freshmeat.net crew.@
05:12 < dooglus> this keyboard changes between US and UK layouts apparently at random!
05:13 < dooglus> thanks for all the testing and bug reports, guys.
05:13 < dooglus> ttyl
05:13 < Zelig> Syea, dooglus
05:14 < pabs3> http://gnomefiles.org/comment.php?soft_id=1167
05:14 < pabs3> later dooglus, thanks heaps
05:15 -!- Zelig [n=Zelig@ip68-108-123-29.lv.lv.cox.net] has left #synfig ["Kopete 0.12.4 : http://kopete.kde.org"]
05:31 < pxegeek> pabs - do you need to update the requirements on gnomefiles?  it lists gtk v. 2.4.x....
05:31 < pxegeek> Not that I think we're specifically calling for any features in newer ones, but if anyones using binary versions....
05:32 < CIA-28> synfig: dooglus * r879 /synfig-core/trunk/src/modules/mod_particle/ (random.cpp random.h): Small changes to make these files more similar to the ones in modules/mod_noise.
05:32 < CIA-28> synfig: dooglus * r880 /synfig-core/trunk/src/modules/mod_noise/ (distort.h noise.cpp random.cpp noise.h random.h distort.cpp): Use an enumeration type rather than unnamed integers to specify the different types of random numbers. Just like r636, but for mod_noise's random numbers. Should we merge the two and move them into synfig-core proper?
05:32 < CIA-28> synfig: dooglus * r881 /synfig-studio/trunk/src/gtkmm/dock_navigator.cpp: 
05:32 < CIA-28> synfig: Clear the dirty flag before starting the render, not after. This shouldn't make
05:32 < CIA-28> synfig: any difference, but if this thread gets swapped out between the two lines and
05:32 < CIA-28> synfig: the rendering finishes before the flag is clearer, we won't see the results of
05:32 < CIA-28> synfig: the render.
05:32 < CIA-28> synfig: dooglus * r882 / (5 files in 2 dirs): Run everything in the same thread.
05:32 < CIA-28> synfig: dooglus * r883 /synfig-studio/trunk/src/gtkmm/asyncrenderer.cpp: Show a message on stdout saying that we're single-threaded.
05:32 < CIA-28> synfig: dooglus * r884 /synfig-studio/trunk/src/gtkmm/workarea.cpp: Only check whether the renderer is updating when single threaded.
05:32 < CIA-28> synfig: dooglus * r885 /synfig-core/trunk/src/modules/mod_noise/noise.cpp: Use 'ftime' instead of 'f' as the floating point time variable to make it easier to search for.
05:32 < CIA-28> synfig: dooglus * r886 /synfig-studio/trunk/src/gtkmm/asyncrenderer.h: Don't use SINGLE_THREADED by default.
05:33 < pabs3> pxegeek: didn't change them
05:34 < pxegeek> ?
05:34 < pabs3> the requirements on gnomefiles
05:34 < pxegeek> I know - should you?
05:34 < pabs3> no idea
05:35 < pxegeek> I guess if you're compiling your own it's no problem
05:35 < pabs3> pxegeek: btw, when you build the new version for windows, please use GTK+ and GTKmm 2.12, and also openexr 1.6.0
05:35 < pxegeek> but if you're trying to match to a precompiled version....
05:35 < pxegeek> OK.
05:40 < pxegeek> Yay - I bumped synfig's rating from 9.13 to 9.16 on Gnomefiles!  Get out and vote!
05:40 < pabs3> :)
05:43 < pxegeek> gtk+ for win seems to be at 2.10....
05:43 < pxegeek> http://gladewin32.sourceforge.net/modules/wfdownloads/visit.php?cid=14&lid=110
05:44 < pabs3> bummer
05:45 < pabs3> we should link here instead of the gladewin32 project I think: http://ftp.gnome.org/pub/gnome/binaries/win32/gtkmm/
05:45 < pxegeek> that's for gtkmm, no?
05:46 < pxegeek> (and we do)
05:46 < pabs3> ah, woops
05:47 < pxegeek> So I think we're running at the latest versions?  Unless we can influence the gtk for windows team to upgrade?
05:47 < pxegeek> I've downloaded Openexr 1.6.0, so assuming that doesn't give me any grief....
05:48 < pabs3> yeah. I guess when they release 2.12 you/atrus can recompile and reupload .07 windows versions
05:48 < pabs3> if it does, it'd be fine to revert to 1.4
05:48 < pxegeek> Sounds like a plan
05:49 < pxegeek> Maybe we should set dooglus on them! 
05:58 < pabs3> :)
06:00 -!- GustoEater [n=doug@cpe-76-169-178-220.socal.res.rr.com] has quit ["Konversation terminated!"]
06:10 -!- zotz [n=zotz@24.244.163.157] has quit ["Leaving"]
07:06 < pxegeek> Well, I think one of the openexr patches is still reqd
07:06 < pxegeek> ('cos it fails without it)
07:07 < pabs3> bummr
07:07 < pxegeek> The other refers to a library that's not in 1.6.0, so that may not be an isasue anymore :)
07:08 < pxegeek> Can I edit patch files by hand?
07:09 < pxegeek> do they care about the stae stamp ofthe files?
07:09 < pxegeek> date stamp?
07:10 < pabs3> date stamp shouldn't matter
07:11 < pabs3> editing the patch isn't recommended, unless you have editdiff
07:14 < pxegeek> Applying patches...
07:14 < pxegeek> patching file `OpenEXR.pc.in'
07:14 < pxegeek> Hunk #1 FAILED at 7.
07:14 < pxegeek> 1 out of 1 hunk FAILED -- saving rejects to OpenEXR.pc.in.rej
07:16 < pabs3> :(
07:16 < pabs3> perhaps that hunk isn't needed any more?
07:17 < pxegeek> No...  they've changed the names of the libraries they're specifyoing
07:17 < pxegeek> and flags
07:17 < pabs3> I'm mailing LWN with a copy of the release announcement I sent them last time
07:17 < pabs3> dooglus: ^^
07:20 < pxegeek> mmm....  think I'm way out of my depth here.  I don't know why Atrus patched it the way he did.  
07:21 < pabs3> can always update it later when gtk win32 is updated
07:23 < pxegeek> well, it's compiling....  let's see if it complains about missing libraries
07:24 -!- ZanQdo [n=Daniel@201.201.2.22] has joined #synfig
07:24 < pxegeek> yup - sure enough....
07:25 < pabs3> synfig or openexr?
07:25 < pxegeek> openexr
07:25 < pabs3> can you paste the errors to pastebin.ca?
07:25 < pxegeek> helpful suggestion from the compiler - Perhaps you should add the directory containing `IlmBase.pc'
07:25 < pxegeek> You think?
07:26 < pxegeek> If I knew which dir that was !
07:33 -!- genete [n=Genete@84.122.44.48.dyn.user.ono.com] has joined #synfig
07:35  * genete has rebuild svn 886 and have errors:
07:36 < genete> when installing synfigstudio
07:36 < genete> asyncrenderer.h:45: error: expected unqualified-id before '<<' token
07:36 < genete> dock_navigator.cpp: In member function 'void studio::Widget_NavView::on_start_render()':
07:36 < genete> dock_navigator.cpp:168: error: invalid use of undefined type 'struct studio::AsyncRenderer'
07:36 < genete> render.h:55: error: forward declaration of 'struct studio::AsyncRenderer'
07:36 < genete> dock_navigator.cpp:169: error: invalid use of undefined type 'struct studio::AsyncRenderer'
07:36 < genete> render.h:55: error: forward declaration of 'struct studio::AsyncRenderer'
07:36 < genete> dock_navigator.cpp:171: error: invalid use of undefined type 'struct studio::AsyncRenderer'
07:36 < genete> render.h:55: error: forward declaration of 'struct studio::AsyncRenderer'
07:36 < genete> /home/carlos/synfig/etl/ETL/_handle.h: In member function 'void etl::handle<T>::detach() [with T = studio::AsyncRenderer]':
07:36 < genete> /home/carlos/synfig/etl/ETL/_handle.h:165:   instantiated from 'etl::handle<T>::~handle() [with T = studio::AsyncRenderer]'
07:36 < genete> dock_navigator.cpp:70:   instantiated from here
07:36 < genete> /home/carlos/synfig/etl/ETL/_handle.h:219: error: invalid use of undefined type 'struct studio::AsyncRenderer'
07:36 < genete> render.h:55: error: forward declaration of 'struct studio::AsyncRenderer'
07:36 < genete> make[2]: *** [synfigstudio-dock_navigator.o] Error 1
07:36 < genete> make[2]: se sale del directorio `/home/carlos/synfig/synfigstudio/src/gtkmm'
07:36 < genete> make[1]: *** [install-recursive] Error 1
07:36 < genete> make[1]: se sale del directorio `/home/carlos/synfig/synfigstudio/src'
07:36 < genete> make: *** [install-recursive] Error 1
07:36 < genete> See you later...
07:37 -!- genete [n=Genete@84.122.44.48.dyn.user.ono.com] has quit [Read error: 104 (Connection reset by peer)]
07:37 < pabs3> genete: you have a conflict, do find -iname asyncrenderer.h | xargs svn revert
07:40 < pxegeek> pabs - http://pastebin.ca/731603
07:41 < pxegeek> THis is without Atrus's patches
07:41 < pabs3> I guess you need http://download.savannah.nongnu.org/releases/openexr/ilmbase-1.0.0.tar.gz installed before openexr
07:44 < pxegeek> That may help....  Looks like you can set a flag to ignore the test?
07:44 < pxegeek> Where would I configure --disable-ilmbasetest
07:49 < pxegeek> Looks like a job for Atrus to me....  
07:49 < pxegeek> It's way past my bedtime, & I have to get up eraly tomorrow. 
07:49  * pxegeek can't even type straight...
07:50 < pxegeek> 'night all
07:54 < pxegeek> So I lied....
07:54 < pxegeek> I had dooglus' single thread version of synfig compiled already
07:55 < pxegeek> Running that appears to fix the 'Adding timeline to pirates/hyperthreading/multicore" bug
07:56 < pxegeek> I have affinity to two CPUs, and can move around the pirate timeline that I added.  YAY DOOGLUS!
07:58 < pxegeek> It also seems much happier with insane clicking on the timeline bar
08:01 < pxegeek> Much happier!  Have to play with this some more, but right now it's looking VERY good!
08:22 -!- pxegeek [n=chatzill@c-71-59-140-184.hsd1.or.comcast.net] has quit [Read error: 113 (No route to host)]
10:42 -!- pabs3 [i=daemon@60.242.186.48] has quit ["Don't rest until all the world is paved in moss and greenery."]
10:51 -!- TMM [n=hp@ip565b35da.direct-adsl.nl] has quit [Read error: 110 (Connection timed out)]
11:28 -!- ZanQdo [n=Daniel@201.201.2.22] has quit ["Adios"]
12:47 -!- rore_ [n=rore@d90-144-57-43.cust.tele2.fr] has joined #synfig
12:48 -!- rore [n=rore@d77-216-244-133.cust.tele2.fr] has quit [Nick collision from services.]
12:48 -!- rore_ is now known as rore
13:05 -!- Ceox [n=Max@dsl-hkigw3-fe2cde00-73.dhcp.inet.fi] has joined #synfig
13:08 -!- rore [n=rore@d90-144-57-43.cust.tele2.fr] has quit [Read error: 110 (Connection timed out)]
13:14 < Ceox> when are the windows binaries available?
13:19 -!- rore [n=rore@d77-218-113-59.cust.tele2.fr] has joined #synfig
13:21 < KiBi> Ceox: As soon as pxegeek finishes them. :)
13:23 < Ceox> ok
13:45 < dooglus> freshmeat: "Your new release (ID 263427) has been stored for verification by the freshmeat.net crew"
13:45 < dooglus> the last one failed
13:46 < dooglus> probably because I gave the address of KiBi's gitweb when it asked for the 'cvsweb' address
14:04 -!- zotz [n=zotz@24.244.163.157] has joined #synfig
14:06 < Ceox> does synfig export to avi?
14:08 -!- Ceox [n=Max@dsl-hkigw3-fe2cde00-73.dhcp.inet.fi] has quit ["Leaving."]
14:11 -!- crazy_bus [n=philip@CPE-124-184-230-3.nsw.bigpond.net.au] has joined #synfig
14:47 -!- dooglus changed the topic of #synfig to: synfig 0.61.07 available from http://www.synfig.com/download/ | http://wiki.synfig.com/ | news: http://www.synfig.com/2007/05/30/synfig-irregular-news/ http://wiki.synfig.com/News | Ask your question and wait - we will answer it eventually!
14:47 < CIA-28> synfig: dooglus * r887 /ETL/tags/stable: Stable Tag: Removing old tag
14:47 < CIA-28> synfig: dooglus * r888 /ETL/tags/stable: Stable Tag: Copying everything over
14:47 < CIA-28> synfig: dooglus * r889 /synfig-core/tags/stable: Stable Tag: Removing old tag
14:47 < CIA-28> synfig: dooglus * r890 /synfig-core/tags/stable: Stable Tag: Copying everything over
14:47 < CIA-28> synfig: dooglus * r891 /synfig-studio/tags/stable: Stable Tag: Removing old tag
14:47 < CIA-28> synfig: dooglus * r892 /synfig-studio/tags/stable: Stable Tag: Copying everything over
15:02 < CIA-28> synfig: dooglus * r893 /ETL/trunk/Makefile.am: Typo.
15:03 -!- crazy_bus [n=philip@CPE-124-184-230-3.nsw.bigpond.net.au] has quit [Remote closed the connection]
15:17 < CIA-28> synfig: dooglus * r894 /ETL/trunk/ETL/_misc.h: Fix rount_to_int() so it rounds to the nearest integer. Previously it was rounding -1.333 to 0.
15:20 -!- zotz [n=zotz@24.244.163.157] has quit ["Leaving"]
15:28 < KiBi> looks like synfig is building fine where the built was attempted.
15:37 < dooglus> cool
15:37 < dooglus> you've uploaded the final release?
15:37 < dooglus> or are you talking about rc3?
15:40 < KiBi> pabs3 uploaded: synfig_0.61.07~rc3-1_i386.changes
16:08 -!- pxegeek [n=chatzill@c-71-59-140-184.hsd1.or.comcast.net] has joined #synfig
16:08 < dooglus> hi pxegeek
16:08 < dooglus> you're early today
16:09 < KiBi> pxegeek: !?!
16:57 < pxegeek> 'morning all
16:58 < pxegeek> I have a conference call with europe :(
16:58 < pxegeek> It's still going on, butnearly finished
17:00 -!- crazy_bus [n=philip@CPE-124-184-230-3.nsw.bigpond.net.au] has joined #synfig
17:00 < pxegeek> But as it's so early , I can take it from home, and have my PC compiling in the background ;)
17:01 < pxegeek> I don't know if I'll get core & studio built & uploaded before I have to go in to the office, but it sounds like there's at least one keen customer....
17:01 < dooglus> :)
17:01 < pxegeek> So dooglus - did you see my preliminary testing of the single thread renderer?
17:02 < dooglus> pxegeek: yes, sounds good!
17:02 < pxegeek> Very good.  
17:02 < dooglus> pxegeek: if you have 2 documents open, both big and slow ones, and zoom in on one, and while it's updating, zoom in on the other,
17:02 < pxegeek> That change didn't make it into 0.61.07 though, right?
17:02 < dooglus> then the first will stop updating until the 2nd has finished
17:03 < dooglus> it didn't, no.  I didn't want to risk breaking 0.61.07
17:03 < dooglus> I committed it to svn right after the release
17:03 < pxegeek> It's just so cool to watch the CPU monitor and see both graphs jump up.
17:03 -!- crazy_bus [n=philip@CPE-124-184-230-3.nsw.bigpond.net.au] has quit [Read error: 104 (Connection reset by peer)]
17:03 < dooglus> both graphs jump up?  really?
17:03 < pxegeek> mmmmm.....  compile not going well....
17:04 < pxegeek> Yup Hyperthreading can do it's thing even on code that's not optimised for threads
17:04 < dooglus> ok
17:04 < pxegeek> its
17:06  * dooglus is currently very confused - I've changed a source file, but the changes just don't make any difference to the running code
17:06 < dooglus> I don't get it
17:06 < pxegeek> Things following the # sign don't affect the code ;)
17:07 < dooglus> :)
17:09 < dooglus> so do both your CPU graphs go to maximum when rendering now?
17:09 < dooglus> and does that mean rendering is twice as fast as before?
17:09 < pxegeek> No, one goes to about 3/4, the other over half...
17:10 < pxegeek> So theoretically slightly faster, but not that you'd notice
17:10 < pxegeek> Ah!  Automake barfed again.  I'll have to hit with a bat tonight...
17:10 < pxegeek> Sorry Ceox!
17:11 < pxegeek> We are currently awaiting a shipment of small, lemon soaped, paper napkins.  Until then, there will be a short delay...
17:12 < pxegeek> Later everyone!
17:13 < dooglus> later!
17:13 < dooglus> return to your seat
17:31 -!- pxegeek [n=chatzill@c-71-59-140-184.hsd1.or.comcast.net] has quit [Read error: 113 (No route to host)]
17:57 < rore> http://sk1.sourceforge.net/ hey, they forgot SIF in their "uniconvertor" project ;] (ok, maybe they never heard of it)
18:00  * KiBi uniconverts some food.
18:02 < CIA-28> synfig: dooglus * r895 /synfig-core/trunk/src/synfig/ (5 files): Use the ETL 'round_to_int' function rather than re-writing it in each of these files.
18:47 < CIA-28> synfig: dooglus * r896 /synfig-core/trunk/src/synfig/layer_pastecanvas.cpp: Fix 1809480 (Zoom in beyond 8.75 in nested canvases fails).
18:49 < dooglus> pxegeek, genete: is 'preview' broken in single-threaded studio for you too?
19:32 < CIA-28> synfig: dooglus * r897 /synfig-studio/trunk/src/gtkmm/preview.cpp: Typo.
20:05 -!- TMM_ [n=hp@ip565b35da.direct-adsl.nl] has joined #synfig
20:06 -!- TMM_ is now known as TMM
20:34 -!- ZanQdo [n=Daniel@201.201.2.22] has joined #synfig
20:38 -!- zotz [n=zotz@24.244.163.157] has joined #synfig
20:47 -!- genete [n=Genete@84.122.44.48.dyn.user.ono.com] has joined #synfig
20:47 < genete> Hi guys.
20:47 < genete> I'm rebuilding again to see if the problem I had yesterday is solved.
20:48 < genete> I'll be out during 30 minutes.
20:48  * genete is out for a while
20:49  * KiBi guesses a while will take 30 minutes.
20:53 -!- zotz [n=zotz@24.244.163.157] has quit ["Leaving"]
20:57  * genete has build and installed core properly... now doing studio
20:57  * genete wanted to say built ;)
21:00 -!- ulrik [n=ulrikbod@81-231-118-204-no53.tbcn.telia.com] has joined #synfig
21:00 < ulrik> hi
21:00 < dooglus> hi genete, ulrik
21:00 < genete> hi dooglus, ulrik 
21:00 < ulrik> hi dooglus, gente
21:00 < ulrik> genete i mean
21:01 < genete> (that mistake is the origin of the name.... cool!)
21:01 < genete> gente = people
21:01 < ulrik> hehe
21:02 < genete> I wrote an email to my girlfriend telling that there were so much people (I wrote "genete") and then Genete becomes a character...
21:02 < genete> I know you know the story.
21:02 < ulrik> ahh, so that's why
21:03 < genete> I was explaining her that there where so many people on a shop I want to buy something (in fact it was a joke because there were nobody...)
21:04 < dooglus> so genete kind of means 'nobody'?
21:04 < genete> and then the confussion make us laught in followings emails and then "Genete" was bonr
21:04 < genete> born*
21:04 < ulrik> well i was cold "bullen" in school and that is not so fancy like gente, it is a simple cake you eat when drinking coffe in sweden
21:05 < genete> dooglus: Not genete is a written mistake of gente == people
21:05 < rore> peopele 
21:05 < genete> (I have to write double = to douglus to make him understand that is is acomparison and not a asignament ...)
21:06 < genete> rore: yeah culd be like that
21:06 < dooglus> genete: heh at the ==
21:06 < genete> ;-)
21:07  * dooglus is adding a checkbox to the gui for 'single threaded'
21:07 < dooglus> so you don't have to have 2 separate builds
21:08 < genete> if gente == people then genete ==peopele else break;
21:08 < genete> dooglus: good improvement
21:24  * genete has built and installed synfigstudio also svn = 897
21:26 < KiBi> FWIW, I pinged a MOTU (Ubuntu guy) to see whether there are interested in pulling the last packages.
21:29 < genete> I've loaded pirates and added a time line and it did not crash but dragging the time cursor is almost insane... I have my computer at 100% since some minutes ago...
21:29 < genete> Also Iv'e checked that the zoom beyond 8.75 bug has been fixed... :)
21:35 < genete> pirates still rendering...
21:37 < genete> pirates has finisehd render... phew!
21:37 -!- genete [n=Genete@84.122.44.48.dyn.user.ono.com] has quit ["Abandonando"]
21:41 < dooglus> pirates is animated??
22:07 -!- xerakko [n=xerakko@debian/developer/xerakko] has joined #synfig
22:09 < ulrik>  been working hard with syncing waypoints and keyrfames and masking...puhh, it's not easy at least not for me, result: http://www.youtube.com/watch?v=yuOvAwqVqu0
22:19 -!- ZanQdo [n=Daniel@201.201.2.22] has quit ["Adios"]
22:29 -!- zotz [n=zotz@24.244.163.157] has joined #synfig
22:57 < ulrik> dooglus: i'm having problems to import .lst files, i made files with .png images and i get this message when synfig crash: synfig(26316): error: Importer::open(): Unknown file type -- png
22:57 < ulrik> synfig(26316): error: Unable to open /Users/ulrikboden/Synfig/synfig-core/examples/dots/dots069.png
22:57 < ulrik> synfig(26316): info: STATE NORMAL: Received Refresh Ducks
22:57 < ulrik> Segmentation fault
22:57 < ulrik> the .lst file with the walk example works fine, but it has jpg files
23:01 < dooglus> ulrik: weird
23:02 < ulrik> yeah, i'll investigate some and we'll see
23:03 < dooglus> ulrik: I'm trying it too
23:03 < ulrik> i'll beat you :) 
23:04 < dooglus> I just loaded 4 .png files from a .lst just fine
23:04 < dooglus> I'm fast :)
23:04 < ulrik> lol you beat me
23:04 < dooglus> my .lst has 5 lines:
23:04 < dooglus> FPS 4
23:04 < dooglus> rotate-and-shrink0001.png
23:04 < dooglus> rotate-and-shrink0002.png
23:04 < dooglus> rotate-and-shrink0004.png
23:04 < dooglus> rotate-and-shrink0005.png
23:05 < ulrik> do you have to have the path in the lst file?
23:05 < ulrik> ahh, i see, i'll try again
23:05 < dooglus> no
23:05 < dooglus> I didn't
23:05 < dooglus> the files, according to identify:
23:05 < dooglus> rotate-and-shrink0001.png PNG 480x480 480x480+0+0 DirectClass 17kb 
23:05 < dooglus> rotate-and-shrink0002.png[1] PNG 480x480 480x480+0+0 DirectClass 15kb 
23:05 < dooglus> rotate-and-shrink0004.png[2] PNG 480x480 480x480+0+0 DirectClass 10kb 
23:05 < dooglus> rotate-and-shrink0005.png[3] PNG 480x480 480x480+0+0 DirectClass 11kb 
23:06 < ulrik> maybe my text editor leave some formated stuff that i don't see?
23:06 < dooglus> I just tried this and it also worked:
23:06 < dooglus> FPS 4
23:06 < dooglus> ~/Desktop/anim/rotate-and-shrink0001.png
23:06 < dooglus> ~/Desktop/anim/rotate-and-shrink0002.png
23:07 < ulrik> what about hte name on the .lst file?
23:08 < dooglus> I just edited the .lst file to have completely wrong image filenames in it - and it still loaded the images OK...
23:08 < dooglus> looks like it only reads the .lst file once, and doesn't notice when you change it.
23:08 < ulrik> arghh....
23:09 -!- omry [n=omry@bzq-84-108-20-56.cablep.bezeqint.net] has quit [Remote closed the connection]
23:09 < dooglus> use a different filename maybe?
23:10 < ulrik> i get this in xterm, synfig(26415): error: Importer::open(): Couldn't find extension
23:10 < ulrik> synfig(26415): error: Unable to open /Users/ulrikboden/Desktop/dots/??FPS 24
23:10 < dooglus> right.  if you open a .lst file you opened before, it doesn't re-read it
23:10 < dooglus> if you use full paths, they don't work
23:11 < ulrik> ok, i'll rename it
23:12 < dooglus> relative path names seem to be OK:
23:12 < dooglus> FPS 4
23:12 < dooglus> pic/rotate-and-shrink0001.png
23:15 < ulrik> hmmm, still no luck
23:15 < dooglus> is it VNC time?
23:20 -!- xerakko [n=xerakko@debian/developer/xerakko] has quit ["Ex-Chat"]
23:23 < ulrik> dooglus: would you mind?
23:23 -!- genete [n=Genete@84.122.44.48.dyn.user.ono.com] has joined #synfig
23:25 < dooglus> ulrik: of course not
23:25 < ulrik> great
23:26 < genete> dooglus: pirates is not animated; I only added a timeline and drag the time cursor. It took sooo long to render.
23:26 < ulrik> my lst file is in Unicode UTF-16, is it right?
23:27  * genete was synchronizing his mobile to its great W98 system. That's the reason he "only" was out an hour and a half...
23:27 < dooglus> ulrik: no, plain ASCII
23:28 < ulrik> dooglus: ahhh, ok maybe that's the problem
23:39 < ulrik> dooglus: now it's working, i used an app "automator", that reads the names of files and save a text file with the names, in some strange way synfig didn't want to read that, when i wrote one my self it worked
23:40 < dooglus> ulrik: try "export LC_ALL=C" before running the app
23:40 < dooglus> to tell it to use a neutral locale
23:41 < dooglus> you'll need to run the app from the same terminal as the export command though
23:47 < CIA-28> synfig: dooglus * r898 /synfig-studio/trunk/src/gtkmm/ (7 files): Use a checkbox in the gui (Toolbox > File > Setup > Misc) for single_threaded, rather than a compile-time #define.
23:51 < KiBi> "LANG=C application" is another common way
23:51 < dooglus> right
--- Log closed Thu Oct 11 00:00:44 2007