Bug 234876

Summary:	Portage tries to build packages known to fail
Product:	Portage Development	Reporter:	Muelli <gentoo-bugs>
Component:	Enhancement/Feature Requests	Assignee:	Portage team <dev-portage>
Status:	RESOLVED DUPLICATE
Severity:	normal
Priority:	High
Version:	unspecified
Hardware:	All
OS:	Linux
Whiteboard:
Package list:		Runtime testing required:	---

Description Muelli 2008-08-16 01:16:08 UTC

I know that =dev-lang/python-2.5.2-r6 is failing for me (bug 234376) and so should portage.

It tries to emerge that package every time and I don't think it's necessary and very annoying. Sure, we shouldn't have failing packages anyway, but the truth is, that they exists.

I think portage should recognize that a package failed, say, three times and don't try to emerge that ebuild anymore. That would save a lot of time and computing power.

This should obviously not be the default as there are many reasons why a build can fail, i.e. running out of disk-space or OOM. So it's not always the packages fault but I feel that this is a handy FEATURE if you know that it's the packages fault.

Reproducible: Always

Comment 1 Andrew Gaffney (RETIRED) gentoo-dev

2008-08-16 02:11:42 UTC

Is there something that's wrong with package.mask? It would fulfill the function you're requesting

Comment 2 Zac Medico gentoo-dev

2008-08-16 02:29:22 UTC


*** This bug has been marked as a duplicate of bug 232086 ***

Comment 3 Muelli 2008-08-16 03:14:10 UTC

Yes, your probably right. I somehow thought that it should do it automatically but not-building packages should be rather seldom so I actuall can do it on my own.

Comment 4 Marius Mauch (RETIRED) gentoo-dev

2008-09-05 15:38:26 UTC

I'd actually consider such a feature harmful (at least for developers and power-users), esp. when testing/debugging new packages where you sometimes _want_ the package fail to build or trying to figure out what exactly is causing it to fail.

Comment 5 Muelli 2008-09-05 16:44:38 UTC

(In reply to comment #4)
> I'd actually consider such a feature harmful (at least for developers and
> power-users), esp. when testing/debugging new packages where you sometimes
> _want_ the package fail to build or trying to figure out what exactly is
> causing it to fail.
> 
Sure, but I'd say, that it's not what the majority of the people using portage want. They, at least I do, want things to just work^tm. And the rationale would be, that if it's known that a package doesn't build, it shouldn't be build! But as I've already mentioned in comment #3, that shouldn't be the case anyway and it's not cumbersome to add it to packages.mask.