On Wed, Dec 18, 2019 at 3:11 PM Paul Smith <psm...@gnu.org> wrote: > I applied your changes although updates were needed. > > For the future, please note that (a) changes should contain updates to the > regression tests to show the error and ensure the fix works, (b) commit > messages need to be formatted properly so that they can be generated into > ChangeLog entries, (c) we cannot use GNU libc-only functions like memrchr() > since GNU make is highly portable and runs on Windows, VMS, etc., not to > mention BSD, MacOS, and other POSIX systems where glibc is not available. > > I added a regression test based on the Savannah bug report, reformatted the > commit messages, and added an implementation of memrchr() for systems where > it doesn't exist.
Thanks for fixing my commit for me! I'll take these points into account next time. Many other libc implementations have memrchr, but honestly, this third commit was just something I noticed when reading the code, and not directly related to the bug I was trying to fix. Your se_implicit test case got left out of your commit. I guess this was unintentional. Presumably, the test case was based on [SV 54161]. The bug can now be closed. Regards, - Jouke
