On Thu, 31 Jan 2013 02:41:28 +0200 Raphael Kubo da Costa <raphael.kubo.da.co...@intel.com> said:
> Carsten Haitzler (The Rasterman) <ras...@rasterman.com> writes: > > > so as a result, evas is following a much stricter clean "init and shtudown > > when u are done" model which at least allows you to write an app, use evas, > > then stop using it and for the next 10 weeks run it without ever using evas > > again and have stuff freed/shut down and out of memory (hopefully). if we > > don't init and shutdown .. then you keep resources around - in practice, > > not a problem, but in theory, can be. > > > > one big nasty though... fc maintains fc cache files. in ~/.fontconfig.... > > files survive a process exit. if any of these files were intended to be > > cleaned up on shutdown (normally are not - but let's assume fc's internals > > change to need to do this) we may see apps leaking junk/files/garbage in > > ~/.fontconfig if we are not letting fc know that t we are done and want to > > delete such out-of-process resources on a "clean" shutdown (yes a segv and > > so on will leave junk... thats life) > > I understand your points. And judging from what you wrote, not calling > FcFini() is a no-go for evas, right? In this case, can you think of > another solution for the problem that doesn't involve changing WebKit > and/or fontconfig? not saying its a no go. i disabled it for years because calling it would cause segv's... this may have been due to other libs doing what you explain... i re-enabled fcfini when i determined this no longer seemed to happen... what i am saying is.. not calling it has side-effects: 1. valgrind bitching. 2. not givin fc a clean message of "i am done" on exit and thus opportunity to clean up temporary out-of-process resources like files (which fc definitely does deal with). in PRACTICE these are not going to be a problem. #1 we just just ignore. #2... is a theoretical problem if fc were to change its behaviour with respect to its files in ~/.fonctonfig it uses for caching. -- ------------- Codito, ergo sum - "I code, therefore I am" -------------- The Rasterman (Carsten Haitzler) ras...@rasterman.com ------------------------------------------------------------------------------ Everyone hates slow websites. So do we. Make your web apps faster with AppDynamics Download AppDynamics Lite for free today: http://p.sf.net/sfu/appdyn_d2d_jan _______________________________________________ enlightenment-devel mailing list enlightenment-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/enlightenment-devel
