Please read this before reporting a bug:
https://wiki.archlinux.org/title/Bug_reporting_guidelines
Do NOT report bugs when a package is just outdated, or it is in the AUR. Use the 'flag out of date' link on the package page, or the Mailing List.
REPEAT: Do NOT report bugs for outdated packages!
https://wiki.archlinux.org/title/Bug_reporting_guidelines
Do NOT report bugs when a package is just outdated, or it is in the AUR. Use the 'flag out of date' link on the package page, or the Mailing List.
REPEAT: Do NOT report bugs for outdated packages!
FS#53068 - [git] PKGBUILD check() misses builddeps, assumes local environment
Attached to Project:
Arch Linux
Opened by Matt Corallo (BlueMatt) - Thursday, 23 February 2017, 19:31 GMT
Last edited by Doug Newgard (Scimmia) - Saturday, 22 April 2017, 17:22 GMT
Opened by Matt Corallo (BlueMatt) - Thursday, 23 February 2017, 19:31 GMT
Last edited by Doug Newgard (Scimmia) - Saturday, 22 April 2017, 17:22 GMT
|
DetailsTo actually pass the test suite git needs at least perl, man, and requires PATH to include /usr/bin/core_perl (for prove). Additionally, it requires that /dev/shm is not mounted with "noexec", which is a common option for /dev/shm.
|
This task depends upon
In any case, the requirement for "/dev/shm" being non-noexec kinda sucks - it prevents systems from auto-building packages in VMs (well, ok, containers and the like) which often do not mount /dev/shm with exec. AFAICT there is relatively little disk write during test, so I'm not sure what the reason is to keep it (also pollutes host state instead of keeping all the build objects in one place).
As for why my builds were previously failing without expect/man-db, I suppose I had something else fucked, because they appear to now, ignore that.