Re: [Pixman] [BUG REPORT] pixman_glyph_cache_create leaks memory that it allocates

2017-09-18 Thread Emil Velikov
On 17 September 2017 at 22:25, Yuri wrote: >> What you reported seems like an user error. Although without a proper >> log nobody can tell you for sure. >> The leak I've spotted is a genuine leak in pixman. > > > Failure to call a destructor in C code could be interpreted as a

Re: [Pixman] [PATCH] Add destructor function to free pixman_implementation_t structs on exit

2017-09-18 Thread Pekka Paalanen
On Mon, 18 Sep 2017 13:46:06 +0300 Siarhei Siamashka wrote: > This fixes a few small memory leaks detected by valgrind. This memory > was allocated once on pixman library load and never freed (but still > was reachable). The fix only helps if the compiler has support

Re: [Pixman] [BUG REPORT] pixman_glyph_cache_create leaks memory that it allocates

2017-09-18 Thread Yuri
On 09/18/17 03:07, Emil Velikov wrote: Failure to call a destructor in C code could be interpreted as a used error, or as a library error. I prefer to call it a library error, because it's easier to just destroy it in the library, and not depend on users. One can call it spaceship if they want

[Pixman] [PATCH] Add destructor function to free pixman_implementation_t structs on exit

2017-09-18 Thread Siarhei Siamashka
This fixes a few small memory leaks detected by valgrind. This memory was allocated once on pixman library load and never freed (but still was reachable). The fix only helps if the compiler has support for __attribute__((constructor)) and __attribute__((destructor)) function attributes.

Re: [Pixman] [BUG REPORT] pixman_glyph_cache_create leaks memory that it allocates

2017-09-18 Thread Pekka Paalanen
On Mon, 18 Sep 2017 03:19:13 -0700 Yuri wrote: > On 09/18/17 03:07, Emil Velikov wrote: > > Yes there's libraries that do all sorts of things to help lazy > > users... I think pixman is not one of them. > > pixman allocates and maintains the object, so it should ensure that it

[Pixman] [PATCH] configure.ac: Fix false GCC vector extensions detection for Clang 4

2017-09-18 Thread Siarhei Siamashka
From: Rob Tsuk Pixman uses __builtin_shuffle builtin function in the code, which relies on GCC vector extensions. So we also need to have it in the configure.ac test snippet code. GCC and Clang have different incompatible builtin functions for this functionality. Clang 4 now

Re: [Pixman] Patch to fix test for GCC vector extensions

2017-09-18 Thread Siarhei Siamashka
On Mon, 18 Sep 2017 07:24:49 -0700 Rob Tsuk wrote: > > On Sep 17, 2017, at 6:40 AM, Siarhei Siamashka > > wrote: > > > > On Mon, 7 Aug 2017 07:52:29 -0700 > > Rob Tsuk > wrote: > > > >> From

Re: [Pixman] [PATCH] Add destructor function to free pixman_implementation_t structs on exit

2017-09-18 Thread Emil Velikov
On 18 September 2017 at 11:46, Siarhei Siamashka wrote: > This fixes a few small memory leaks detected by valgrind. This memory > was allocated once on pixman library load and never freed (but still > was reachable). The fix only helps if the compiler has support for

Re: [Pixman] Patch to fix test for GCC vector extensions

2017-09-18 Thread Rob Tsuk
Your proposed modification to my patch makes sense. I’m not certain when I”ll have time to perform it. > On Sep 17, 2017, at 6:40 AM, Siarhei Siamashka > wrote: > > On Mon, 7 Aug 2017 07:52:29 -0700 > Rob Tsuk > wrote: > >>