Please support arbitrary arch/buildd affinity for arch:all builds

Bug #217427 reported by Adam Conrad
This bug affects 4 people
Affects Status Importance Assigned to Milestone
Launchpad itself
William Grant

Bug Description

There are some rare cases (palo and openhackware are the current ones) where an arch:all package contains arch-specific compiled code (palo is an hppa bootloader, but arch:all to support cd image mastering, openhackware is powerpc firmware code, but arch:all so it can run in arch:any powerpc emulators).

It would be handy if, on an as-needed basis, I could say "please ignore the arch:all = i386 rule for this package, and build it on hppa/powerpc instead".

Related branches

Revision history for this message
James Troup (elmo) wrote :

This is pretty important (despite the apparently limited number of packages involved) since the current workaround is basically to upload binaries via the backdoor which is evil and wrong in so many (hopefully obvious) ways.

Changed in soyuz:
status: New → Confirmed
William Grant (wgrant)
tags: added: motu
Celso Providelo (cprov)
Changed in soyuz:
assignee: nobody → Celso Providelo (cprov)
importance: Undecided → High
status: Confirmed → Triaged
tags: added: soyuz-build
Revision history for this message
Loïc Minier (lool) wrote :

This is still an issue for packages such as openhackware (bug #64501) or openbiios-sparc (bug #183495)

Changed in soyuz:
assignee: Celso Providelo (cprov) → nobody
Revision history for this message
Julian Edwards (julian-edwards) wrote :

I chatted to Persia yesterday and we can probably add a new header like

X-Arch-Indep-Build-Arch: sparc

and use that to override the nominated arch-indep. Making Soyuz dispatch builds to any arch for arch:all is a much harder fix.

Revision history for this message
Julian Edwards (julian-edwards) wrote :

If someone wants to attempt a fix on this I'm happy to mentor.

tags: added: feature
Revision history for this message
Trung Ngô (ndtrung4419) wrote :

Some one please stand up and fix this. As you can see in the link below, the i386 buildbots are too hard at work while all other bots stand idle, which is an IMMENSE WASTE. Whenever I put an "any" package on my ppa, the amd64 build happens in less than 5 mins while the i386 build takes forever.

Revision history for this message
Colin Watson (cjwatson) wrote :

@Ngô Trung: I'm afraid that has nothing to do with this bug report, despite some superficial resemblance. That's more like bug 285206.

William Grant (wgrant)
Changed in launchpad:
assignee: nobody → William Grant (wgrant)
status: Triaged → In Progress
Revision history for this message
Launchpad QA Bot (lpqabot) wrote :
tags: added: qa-needstesting
Changed in launchpad:
status: In Progress → Fix Committed
William Grant (wgrant)
tags: added: qa-ok
removed: qa-needstesting
Revision history for this message
William Grant (wgrant) wrote :

Source packages can now specify a Build-Indep-Architecture field containing an architecture spec on which to build the architecture-independent packages.

This isn't just a simple override of nominatedarchindep. Arch-indep will only ever be built on an architecture matching the indep spec, and a build will be created if necessary and possible even if Build-Indep-Architecture's value is mutually exclusive with Architecture.

One corner case when the field is omitted has changed: "all" now only forces a nominatedarchindep build if it is the sole term in the Architecture field. Previously an Architecture value of "doesnotexist all" would resolve to nominatedarchindep, but that must now be forced with a Build-Indep-Architecture.

Colin Watson (cjwatson)
Changed in launchpad:
status: Fix Committed → Fix Released
To post a comment you must log in.
This report contains Public information  Edit
Everyone can see this information.

Other bug subscribers