Yuri, please don't use HTML emails. It completely messes up the quotation. On 17 September 2017 at 21:17, Yuri <y...@rawbw.com> wrote: > On 09/17/17 13:07, Emil Velikov wrote: > >> Having the opposite - a destructor [1] should provide symmetry and >> consistency. >> Furthermore using atexit is not as portable/reliable as one would think. >> > > It should be the destructor that handles this. Either the user clears it, or > destructor clears it. > I don't think it's pixman's job to hold the user's hand. If the user does not clear what it creates, then the user should be fixed. As mentioned - pixman emits lovely BUG notations when that happens.
>> All this is obviously orthogonal to the original issue reported ;-) > > > I don't see how it is orthogonal. > 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. > > There is actually __attribute__((destructor)) > https://phoxis.org/2011/04/27/c-language-constructors-and-destructors-with-gcc > It works with gcc and clang, and probably with most or all other compilers. > This is precisely what I recommended, haven't I? -Emil _______________________________________________ Pixman mailing list Pixman@lists.freedesktop.org https://lists.freedesktop.org/mailman/listinfo/pixman