On Fri, 29 Jul 2011 12:21:10 +0100 Michael Meeks <michael.meeks@novell.com> wrote:
Which reminds me; why are we not checking in configure for make 2.82 ? IIRC there was some to/fro on that - but I forget the result; not checking for a make with the fix we need is a tad odd right ?
Yes, there seems to be a hard to reproduce error in variable expansion in gnu make 3.82. I gitified the gnu make cvs and created the attached script for bisecting. However, I am not even sure I cant trigger the error with that. One run of gnu make test on my tree takes ~40 secs, so 100 runs are ~1 hours. There are 115 revisions between 3.81 and 3.82 so it should take about 7 hours (if the test is reliable) to find the offending change. From that I hope it would be rather easy. Best, Bjoern -- https://launchpad.net/~bjoern-michaelsen
Attachment:
bisectmake
Description: Binary data