Chris Lamb a écrit : > ... > Your trick can cause more subtle problems though - if the original > tarball is not found then an entirely different clean routine is used, > which leads to builds that are not easy to reproduce. This is compounded > by your brittle calculation of UPSTREAMVERSION which does not correctly > calculate the upstream version component in all cases (eg. binNMUs).
OK, this is understandable. > [...] Anyway, the fix is fairly simple; just remove all the tarball > messing-about and expand your clean target to clean up after upstream. I shall do it. The reason why I did not this before is that the upstream sources were never intended to be properly cleaned before last year. Making a list of all the failures in the clean target was fragile, since the list of files might vary along versions. There used to be a single author for all the package, which gave no right to contributors to commit changes directly. Now this organization has changed, and the upstream sources are managed in a SNV public repository, where I can have a write access. There is much to do in the beta version, and my work will be done taking in account your suggestion there. Best regards, Georges.
signature.asc
Description: Digital signature
