Re: Are soname bumps required when library upgrades break compatibility?
Policy 8.1 is clear - if the bug documents a crash in an application that was not present before the library was updated and the library has not changed the SONAME or package name, the library justifies an RC bug. Actually, Policy 8.1 doesn't say anything about when to change the soname. Perhaps it should? The bit about the SONAME is implicit in the change of package name (as checked by lintian) and if the package name has changed, then the application will have had to have been rebuilt against the new library API so a crash in those circumstances is still a bug, this time caused by a buggy library AFTER a correct transition. I was just pointing out that if the transition is correct, the bug does not have to be RC. Policy mandates that the library must transition cleanly - the SONAME is just the mechanism used to change the package name which is what actually determines what gets installed. Bugs which result from an updated library package being installed when it should have been held back are examples of an incorrect library transition and break Policy 8.1. There is no need for Policy to delve into the mechanism of complying with Policy as long as the result of the changes has the effect of compliance. -- Neil Williams = http://www.data-freedom.org/ http://www.nosoftwarepatents.com/ http://www.linux.codehelp.co.uk/ pgpF1J19eic6N.pgp Description: PGP signature
Re: Are soname bumps required when library upgrades break compatibility?
Brandon [EMAIL PROTECTED] (11/09/2007): Yes, serious at least, but I'd even say “grave”: “renders package unusable” (by its dependencies) in reportbug. That is a bit of a stretch. It would not be unusable. The point of a library is to be used by its dependencies, by the packages built against it. If it is (not|no longer) usable by these rdepends, the package is unusable. You can't really say breaks unrelated packages either, because they would have a direct relation. I didn't say anything like that. I was just wondering about how I would justify keeping the bug at RC if the maintainer wanted to downgrade it. He is cooperative, and is genuinely trying his best to maintain a good stable package, but I like to plan ahead. I also thought it was odd that soname was barely mentioned in debian policy, as it is so important. That may explain why libpkg-guide exists. The policy is a more general document. I recommended to Patrick, the maintainer of guichan, that he writes to this list to ask about how he should handle his package. Thanks for that. Cheers, -- Cyril Brulebois pgpVqAZPtg3jc.pgp Description: PGP signature
Re: Are soname bumps required when library upgrades break compatibility?
That is a bit of a stretch. It would not be unusable. The point of a library is to be used by its dependencies, by the packages built against it. If it is (not|no longer) usable by these rdepends, the package is unusable. That makes sense. You can't really say breaks unrelated packages either, because they would have a direct relation. I didn't say anything like that. I didn't mean to imply that you did. It was actually something I thought of. Your justification makes more sense. Policy 8.1 is clear - if the bug documents a crash in an application that was not present before the library was updated and the library has not changed the SONAME or package name, the library justifies an RC bug. Actually, Policy 8.1 doesn't say anything about when to change the soname. Perhaps it should? Thanks for your explanations guys. I get it now. A crash is serious, whether or not the reason is documented in policy. If the crash is the fault of the library, the library gets the RC bug. The maintainer should not downgrade it and you would be justified in reinstating that severity. Ok. I also wanted to make it clear that Patrick never tried to downgrade the bug. I just wanted to be prepared in case some maintainer I ever need to deal with in the future did. -Brandon -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Are soname bumps required when library upgrades break compatibility?
On Tue, Sep 11, 2007 at 06:27:11PM -0700, Brandon wrote: Thanks for your explanations guys. I get it now. A crash is serious, whether or not the reason is documented in policy. If the crash is the fault of the library, the library gets the RC bug. The statement was that a crash due to changes in a dependency is a severity:serious bug in that dependency. A crash is always a bug, but many are just severity:normal for non-core functionality or important for things that don't totally inhibit the package's utility. Justin -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]