Ben Elliston wrote: > > You are assuming that system programs have a test suite that will > > protect them from introducing regressions? I don't think that's realistic. > > Even glibc has regressions occasionally. > > Not really. It is not difficult for programmers to be disciplined and > know then they are breaking an interface. Why did they do this?
It looks like they did not think at the 'ldd --version' interface [1][2]. Nor did they backport their own fix [3] to the 3.10-stable branch. > > What do you hope to achieve by refusing to fix this in config.guess? > > My aim is to keep config.guess maintainable. Is it really the job of > config.guess to patch up everyone else's mistakes? I share your feeling... But workarounds are a fact of life. Bruno [1] https://lists.alpinelinux.org/~alpine/devel/%3CC0FBEB97-17F6-4B0B-B715-0B06EFAA062D%40bernhard-ehlers.de%3E [2] https://git.alpinelinux.org/aports/commit/?h=3.10-stable&id=d4a7955cc72b9395779c20755df88f983763c43b [3] https://git.alpinelinux.org/aports/commit/?h=3.11-stable&id=2f2f995d9d52484720dbe8b4b1649c132ec1feee