On 31/12/2020 09:20, Roger Pau Monné wrote: > On Wed, Dec 30, 2020 at 06:32:58PM +0000, Andrew Cooper wrote: >> On 30/12/2020 17:34, Roger Pau Monne wrote: >>> It's not part of the POSIX standard [0] and as such non GNU ar >>> implementations don't usually have it. >>> >>> It's not relevant for the use case here anyway, as the archive file is >>> recreated every time due to the rm invocation before the ar call. No >>> file name matching should happen so matching using the full path name >>> or a relative one should yield the same result. >>> >>> This fixes the build on FreeBSD. >>> >>> Signed-off-by: Roger Pau Monné <[email protected]> >> Acked-by: Andrew Cooper <[email protected]>, although... >> >> We really need some kind of BSD build in CI. This kind of breakage >> shouldn't get into master to begin with. > Fully agree. I'm not that familiar with gitlab CI, but since we have > our own runners there, could we also launch VMs instead of just using > containers? > > It might be easier to integrate with osstest in the future, since > FreeBSD has now switched to git. > >>> [0] https://pubs.opengroup.org/onlinepubs/9699919799/utilities/ar.html >>> --- >>> I'm unsure whether the r and s options are also needed, since they >>> seem to only be relevant when updating a library, and Xen build system >>> always removes the old library prior to any ar call. >> ... I think r should be dropped, because we're not replacing any files. >> However, I expect the index file is still relevant, because without it, >> you've got to perform an O(n) search through the archive to find a file. > Right, the description of 's' between opengroup and the Linux man page > seems to be slightly different. From the opengroup description it seems > like s should be used to force the generation of a symbol table when > no changes are made to the archive, but otherwise ar should already > generate such symbol table.
Ok - are you happy for me to commit with dropping the r and s? ~Andrew
