Deprecating aspell package because there are better-supported spell checkers like hunspell/enchant2 which could be used instead. It also has an upstream with almost 4 years of no action.
Owners, do not implement this work until the FESCo vote has explicitly ended. The Fedora Program Manager will create a tracking bug in Bugzilla for this Change, which is your indication to proceed. See the FESCo ticket policy and the Changes policy for more information.
REMINDER: This ticket is for FESCo members to vote on the proposal. Further discussion should happen in the devel list thread linked above.
Pet peeve: Deprecation vs. Depreciation ... The former has a technical meaning in the context of Fedora (i.e. "officially frown upon using it") , the second one just means "making worth less" ...
+1 (and +1 to @decathorpe's peeve)
The peeve is only in this issue, the change is correctly worded
me edits the issue title :)
forgot to vote +1
Well, one could argue that if Fedora deprecates something, it depreciates in value universally :smile_cat:
+1
How confident can we be that hunspell is going to outlive aspell in the long term? The hunspell project has been more active recently, but it went four years without a patch release from 2018 to 2022. In 2020 or so, aspell might have looked like the more active project.
I'm open to orphaning the package if deprecating it would be risky.
+1 here
Why not both? Deprecation does not imply immediate removal. I.e. you could mark the package as deprecated to make sure no more software that uses aspell is packaged for Fedora, but people could still step up to continue maintaining it for dependent packages that are already in Fedora.
I'm open to orphaning the package if deprecating it would be risky. Why not both? Deprecation does not imply immediate removal. I.e. you could mark the package as deprecated to make sure no more software that uses aspell is packaged for Fedora, but people could still step up to continue maintaining it for dependent packages that are already in Fedora.
Yes, that's even better
Not sure it needs to be deprecated. Slow updates often occur for software written in programming languages that are not high maintenance. Announcing intent to orphan it seems fine, allows others who are interested in it to take up maintenance.
Has a formal decision been reached for this change proposal?
APPROVED (+5, 0, 0)
Metadata Update from @zbyszek: - Issue tagged with: pending announcement
Thank you, will process this change now
Hi,
I was wondering if I can deprecate also all of the aspell dictionaries (aspell-* packages).
There is one issue, I'm not a maintainer of all of them, so I would use the proven-packager permissions. I will let the maintainers know in the form of Bugzilla, but that might be inactive. I would suggest waiting for 7 days, and if there is no response, proceed with deprecating them.
Is it okay with you?
Have an open review for Aspell dictionaries for Kiswahili. My understanding is that Finnish and a few other languages use Aspell dictionaries. Would be happy to pick up the package unless there are security concerns. Orphaning would have been better, but can move them to copr.
The aspell package has been already deprecated, I don't see any reason why its dictionaries (packages that don't work without aspell package) shouldn't be. Deprecation won't change the packages in any way, it won't remove them either. I will probably also orphan the packages, so once orphaned feel free to take them.
@ljavorsk I think not many maintainers knows how to deprecate Fedora package but know how to retire or orphan them. You may want to guide on opened bugzilla, how maintainers can deprecate package.
Thanks for the tip, I've provided the links and info about how to deprecate a package (like this: https://bugzilla.redhat.com/show_bug.cgi?id=2218586#c1)
This package should not have been deprecated. It should only have been orphaned as the package still builds, does not seem to present a security concern and can be useful. There does not seem to be a policy for package deprecation, though there is one for orphaning and retirement: https://docs.fedoraproject.org/en-US/fesco/Policy_for_orphan_and_retired_packages/#_orphaning_and_retiring_packages Probably a policy is also needed for packages where patches may be needed and are provided by the packager, but upstream is no longer active.
Proposed policy change https://fedoraproject.org/wiki/Changes/DeprecationPolicy#Upgrade%2Fcompatibility_impact
There's already a policy for deprecating packages, in the packaging guidelines: https://docs.fedoraproject.org/en-US/packaging-guidelines/deprecating-packages/
That policy does not give reasons for when one should deprecate as opposed to orphan or retire.
This broke hunspell-en dictionary. The change in https://src.fedoraproject.org/rpms/hunspell-en/c/a5d6fd95d1a110f9fc9fdcc7c8f2d53269f3f137?branch=rawhide which removed BR: aspell made the dictionary generation script unusable, as there is no support for building using hunspell tools. And this is causing other packages build failing, see https://bugzilla.redhat.com/show_bug.cgi?id=2218460
BR: aspell
So, is someone willing to propose a patch upstream at https://github.com/en-wl/wordlist or has any other solution?
Is that why Firefox thinks everything I type is spelled wrong now? I don't know exactly when this started, but it's rather annoying. If I right-click and change the language from "English (Canada)" to "English (United Kingdom)" or "English (United States)" it seems to fix it, but I can't really do that for every box I ever type anything into in Firefox...
Hi, @mattia I don't think it's good to remove aspell dependency without having it enabled in upstream. By deprecating aspell we didn't force other packages to remove aspell dependency, we just recommended them to either remove the dependency if possible, or start the conversation with upstream towards this goal. Bottom line: you can still use aspell, if needed.
Announced: https://lists.fedoraproject.org/archives/list/devel@lists.fedoraproject.org/message/64JVFPZ7BS3PNGAIQTRKSWNZEB6CZ7QD/
I'll leave it open because of the ongoing discussion.
Metadata Update from @zbyszek: - Issue untagged with: pending announcement
After reading https://bugzilla.redhat.com/show_bug.cgi?id=2218460, I think the discussion here is mostly based on a misunderstanding. There was some bug in hunspell that was later fixed (maybe only partially, see https://bugzilla.redhat.com/show_bug.cgi?id=2158548, https://github.com/hunspell/hunspell/issues/903). This isn't directly related to whether aspell is deprecated. The whole point of marking aspell is deprecated is for packages to move over to other changes on their own schedule.
@fed500, since you have a package review open for another aspell dict, please consider changing it to hunspell or one of the alternatives. It's better overall to agree on one or a few checkers. If it turns out to be impossible to move away from aspell, you can always discuss with @ljavorsk taking over the maintenance of the package and undeprecating it, but it seems to early for this. Let's try to complete the move first.
I'll close this, since the deprecation happened as planned.
Metadata Update from @zbyszek: - Issue close_status updated to: Accepted - Issue status updated to: Closed (was: Open)
Log in to comment on this ticket.