Summary: | packages masked by nothing | ||
---|---|---|---|
Product: | Portage Development | Reporter: | Marijn Schouten (RETIRED) <hkbst> |
Component: | Unclassified | Assignee: | Portage team <dev-portage> |
Status: | RESOLVED FIXED | ||
Severity: | normal | CC: | ferringb, jakub, masterdriverz |
Priority: | High | Keywords: | InVCS |
Version: | 2.1 | ||
Hardware: | AMD64 | ||
OS: | Linux | ||
Whiteboard: | |||
Package list: | Runtime testing required: | --- | |
Bug Depends on: | |||
Bug Blocks: | 147007 |
Description
Marijn Schouten (RETIRED)
2006-10-09 08:24:36 UTC
This is not a bug... sys-kernel/vanilla-sources -~amd64 in package.keywords + ~amd64 in ebuild keywords produces masked by "nothing", since you've "nullified" the keywords. All other versions are package.masked, so there's nothing available. #only accept stable vanilla-sources: sys-kernel/vanilla-sources -~amd64 #accept unstable vanilla-sources-2.6.18: =sys-kernel/vanilla-sources-2.6.18 Combined with ~amd64 keyword in vanilla-sources-2.6.18 ebuild should result in an acceptable vanilla-sources-2.6.18, or at least in an unacceptable vanilla-sources-2.6.18 which is masked by ~amd64. But masked by nothing? No. -~amd64 is a strange kind of syntax to say at least. What exactly do you want to express with that? (In reply to comment #3) > -~amd64 is a strange kind of syntax to say at least. What exactly do you want > to express with that? man portage <snip> Example: # only use stable mplayer media-video/mplayer -~x86 </snip> Well, it overrides all ~arch keywords, so subsequently accepting ~arch for a particular version has no effect. <sys-kernel/vanilla-sources-2.6.18 -~amd64 ~sys-kernel/vanilla-sources-2.6.18 >sys-kernel/vanilla-sources-2.6.18 -~amd64 in package.keywords does what the reporter expects. In svn r5121, I've stacked package.keywords incrementally so that -keyword works as expected for any incremental variable. This brings the gvisible and getmaskingstatus logic into sync so that the confusing (masked by: ) message will not be generated in cases where package.keywords contains -keyword followed by keyword. (In reply to comment #5) > In svn r5121, I've stacked package.keywords incrementally so that -keyword > works as expected for any incremental variable. This brings the gvisible and > getmaskingstatus logic into sync so that the confusing (masked by: ) message > will not be generated in cases where package.keywords contains -keyword > followed by keyword. You do realize that the only way to match -arch (which is stupid in keywords, but there) is via -* now, right? (In reply to comment #6) > You do realize that the only way to match -arch (which is stupid in keywords, > but there) is via -* now, right? Now I know. Thanks. :) In svn r5123 it's fixed to allow those to be matched. This has been released in 2.1.2_rc2-r2. |