On Tue, Jun 05, 2012 at 11:35:53AM +0300, Konstantin Belousov wrote: > Not to mention that the patch was committed to _our_ implementation of > libc, which uses _our_ free, and not some abstract free(3). Our free changing > errno means that process state is so messed that worrying about realpath(3) > correctness is beyond any expectations.
We can't stay with "our free" concept but with "some abstract free" instead. The code must be portable and even "our free" can be replaced in the future. About errno changing, look at "our free" code pass, potential candidates (as Garrett mentions) are utrace, assert, idalloc - I don't look deep to say for sure. But it does not really matter because the whole "our free" concept is flawed. > I already described my POV to ache, but it seems that nobody listens. It just > a season, it seems. You deside to not answer to the free() discussion which follows your initial commit, which means either you agree with change or you are not interested. It looks strange that you just consider to answer now and keep silence at the time of discussion. -- http://ache.vniz.net/
pgpgaVwvf3Xir.pgp
Description: PGP signature
