Summary: | emerge --fetchonly should not require USE dependencies to be satisfied | ||
---|---|---|---|
Product: | Portage Development | Reporter: | Nick Bowler <nbowler> |
Component: | Core - Interface (emerge) | Assignee: | Portage team <dev-portage> |
Status: | RESOLVED WONTFIX | ||
Severity: | enhancement | CC: | fedja, jaglover, sam |
Priority: | High | ||
Version: | unspecified | ||
Hardware: | All | ||
OS: | Linux | ||
Whiteboard: | |||
Package list: | Runtime testing required: | --- | |
Bug Depends on: | |||
Bug Blocks: | 377365 |
Description
Nick Bowler
2010-07-20 18:28:55 UTC
No, it shouldn't. The current hack for unsatisfied deps already leads to strange behavior (bug 314579). The resulting fetch could be incomplete because use flags change which deps are pulled in, not to speak of use conditionals in SRC_URI. (In reply to comment #1) > No, it shouldn't. The current hack for unsatisfied deps already leads to > strange behavior (bug 314579). That looks like an unrelated bug to me -- it looks like too much stuff gets fetched there? IMO it's totally fine if too much stuff gets fetched. > The resulting fetch could be incomplete because use flags change which deps > are pulled in, not to speak of use conditionals in SRC_URI. Conditionals in SRC_URI should not be a problem in the --fetch-all-uri case. (In reply to comment #2) > (In reply to comment #1) > > No, it shouldn't. The current hack for unsatisfied deps already leads to > > strange behavior (bug 314579). > > That looks like an unrelated bug to me -- it looks like too much stuff gets > fetched there? IMO it's totally fine if too much stuff gets fetched. > > > The resulting fetch could be incomplete because use flags change which deps > > are pulled in, not to speak of use conditionals in SRC_URI. > > Conditionals in SRC_URI should not be a problem in the --fetch-all-uri case. > Especially in the USE case, it could be the other way around. Anyway, people will always complain. I have the same problem on my server, which fetches files needed for my Gentoo boxes during the night. The list of packages on the Gentoo boxes is created by equery list -F '\$cp' '*' an then copied to the server. The server of course has different use flags than my Gentoo boxes. I even tried with --keep-going like: "emerge --fetch-all-uri --keep-going `cat /tmp/all-installed-pkgs`" but it didn't work either. So maybe --keep-going (in combination with --fetchonly|--fetch-all-uri) should be extended so that it skips unmet requirements because in this case the user told emerge via --keep-going that it's fine if not all of the packages could be processed. (In reply to comment #4) > I have the same problem on my server, which fetches files needed for my Gentoo > boxes during the night. > The list of packages on the Gentoo boxes is created by > equery list -F '\$cp' '*' > an then copied to the server. > > The server of course has different use flags than my Gentoo boxes. I think this goes beyond the scope of what emerge --fetch should do. I've filed bug 382089 for a more general purpose mirroring tool. (In reply to comment #1) > The resulting fetch could be incomplete because use flags change which deps are > pulled in, not to speak of use conditionals in SRC_URI. Right. It's just not reliable enough to be worth the trouble, so I'm closing this as WONTFIX. We can consider adding emerge --fetch like options to the mirroring tool that's planned for bug 382089. *** Bug 890303 has been marked as a duplicate of this bug. *** *** Bug 533130 has been marked as a duplicate of this bug. *** |