[aur-requests] [PRQ#11354] Deletion Request for firefox-always-nightly
Morganamilo [1] filed a deletion request for firefox-always-nightly [2]: Duplicate of aur/firefox-nightly See https://lists.archlinux.org/pipermail/aur- general/2018-May/034080.html [1] https://aur.archlinux.org/account/Morganamilo/ [2] https://aur.archlinux.org/pkgbase/firefox-always-nightly/
How do I tell this request to, politely, go away? always-nightly has a number of differences to -nightly, not just the "always" part of it. But if this *is* the main point of difference, then I contest the claim that -nightly has "pretty good tracking". As a blatant example, right now it is on 62.0a1.201805*09*, while always-nightly downloads 62.0a1.201805*10**.* There is also the fact that firefox nightly updates *twice a day*, not once, so the -nightly package is *never* accurately tracking the latest. Some more occurences from the firefox-nightly comments: owstoni <https://aur.archlinux.org/account/owstoni> commented on 2017-08-02 22:15 57 is out, could you post the new source? K900 <https://aur.archlinux.org/account/K900> commented on 2017-08-02 14:40 57.0a1 is out now, needs a version bump. tux-86 <https://aur.archlinux.org/account/tux-86> commented on 2017-07-21 20:11 Version string seems now 56.0a1.-1 That's two fairly recent occasions where firefox-nightly was behind. The 57 one shows no update in 8 hours at least. Firefox-always-nightly would have produced the right version within a few minutes of the update actually going out. *That* is why it exists. On Fri, May 11, 2018 at 9:15 AM <notify@aur.archlinux.org> wrote:
Morganamilo [1] filed a deletion request for firefox-always-nightly [2]:
Duplicate of aur/firefox-nightly
See https://lists.archlinux.org/pipermail/aur- general/2018-May/034080.html
[1] https://aur.archlinux.org/account/Morganamilo/ [2] https://aur.archlinux.org/pkgbase/firefox-always-nightly/
On 10 May 2018 at 22:36, Félix Saparelli <felix@passcod.name> wrote:
How do I tell this request to, politely, go away?
always-nightly has a number of differences to -nightly, not just the "always" part of it. But if this is the main point of difference, then I contest the claim that -nightly has "pretty good tracking". As a blatant example, right now it is on 62.0a1.20180509, while always-nightly downloads 62.0a1.20180510.
There is also the fact that firefox nightly updates twice a day, not once, so the -nightly package is never accurately tracking the latest.
Some more occurences from the firefox-nightly comments:
owstoni commented on 2017-08-02 22:15
57 is out, could you post the new source?
K900 commented on 2017-08-02 14:40
57.0a1 is out now, needs a version bump.
tux-86 commented on 2017-07-21 20:11
Version string seems now 56.0a1.-1
That's two fairly recent occasions where firefox-nightly was behind. The 57 one shows no update in 8 hours at least. Firefox-always-nightly would have produced the right version within a few minutes of the update actually going out. That is why it exists.
On Fri, May 11, 2018 at 9:15 AM <notify@aur.archlinux.org> wrote:
Morganamilo [1] filed a deletion request for firefox-always-nightly [2]:
Duplicate of aur/firefox-nightly
See https://lists.archlinux.org/pipermail/aur- general/2018-May/034080.html
[1] https://aur.archlinux.org/account/Morganamilo/ [2] https://aur.archlinux.org/pkgbase/firefox-always-nightly/
I would think the rules on this would say it's still a duplicate. If there's problems they should be fixed, not have a new package pop up. Your PKGBUILD does a lot of hacky stuff that I don't know how to feel about it. using curl for global variables. Forcing an insanely high pkgver so that it always wants to update. It's not up to me though. I brought this up briefly in the linked thread and Eli seemed to agree so I thought I would send this request and see what happens.
Yeah, I'm deliberately in contravention of the rules and style of AUR to get those advantages. Also note it's not really a "new" package (four years)! But I understand if it gets deleted. Although, if it does get picked for deletion is there a way I could request a stay of execution? I'm investigating an alternate solution that wouldn't rely on the AUR at all, and therefore wouldn't push against its rules, but it's going to take a while (~1 month?) to get set up, and if I could offer a good transition strategy to those who like this flavour, then voluntarily kill it, that would probably be a better user experience. On Fri, May 11, 2018 at 9:49 AM Morgan Adamiec <morganamilo@gmail.com> wrote:
On 10 May 2018 at 22:36, Félix Saparelli <felix@passcod.name> wrote:
How do I tell this request to, politely, go away?
always-nightly has a number of differences to -nightly, not just the "always" part of it. But if this is the main point of difference, then I contest the claim that -nightly has "pretty good tracking". As a blatant example, right now it is on 62.0a1.20180509, while always-nightly downloads 62.0a1.20180510.
There is also the fact that firefox nightly updates twice a day, not once, so the -nightly package is never accurately tracking the latest.
Some more occurences from the firefox-nightly comments:
owstoni commented on 2017-08-02 22:15
57 is out, could you post the new source?
K900 commented on 2017-08-02 14:40
57.0a1 is out now, needs a version bump.
tux-86 commented on 2017-07-21 20:11
Version string seems now 56.0a1.-1
That's two fairly recent occasions where firefox-nightly was behind. The 57 one shows no update in 8 hours at least. Firefox-always-nightly would have produced the right version within a few minutes of the update actually going out. That is why it exists.
On Fri, May 11, 2018 at 9:15 AM <notify@aur.archlinux.org> wrote:
Morganamilo [1] filed a deletion request for firefox-always-nightly [2]:
Duplicate of aur/firefox-nightly
See https://lists.archlinux.org/pipermail/aur- general/2018-May/034080.html
[1] https://aur.archlinux.org/account/Morganamilo/ [2] https://aur.archlinux.org/pkgbase/firefox-always-nightly/
I would think the rules on this would say it's still a duplicate. If there's problems they should be fixed, not have a new package pop up. Your PKGBUILD does a lot of hacky stuff that I don't know how to feel about it. using curl for global variables. Forcing an insanely high pkgver so that it always wants to update.
It's not up to me though. I brought this up briefly in the linked thread and Eli seemed to agree so I thought I would send this request and see what happens.
On 05/10/2018 05:49 PM, Morgan Adamiec via aur-requests wrote:
I would think the rules on this would say it's still a duplicate. If there's problems they should be fixed, not have a new package pop up.
This. With two packages, the only result is no one is quite sure which one to install, because they both seem to be the same package. "always" is not exactly a clear descriptor...
Your PKGBUILD does a lot of hacky stuff that I don't know how to feel about it. using curl for global variables. Forcing an insanely high pkgver so that it always wants to update.
I've seen worse than the curl. The insanely high pkgver is honestly a lot more bothersome. It's flat-out wrong. The technical solution to this is to use an AUR helper which treats "-nightly" as a devel package like "-git" or "-hg".
It's not up to me though. I brought this up briefly in the linked thread and Eli seemed to agree so I thought I would send this request and see what happens.
Thanks. Given the response from Felix in the comments is: ``` Hi all. This package is flagged for deletion, and for very valid reasons: basically it's doing a lot of very hacky stuff to achieve its goals, and that puts it squarely against the rules of the AUR. (I knew this from day 1! it's not a new concern. But I didn't really have users who relied on this package for the longest time, so it didn't matter as much.) I think I have a better solution, but I might not get there in time. Your options are: use firefox-nightly, which most of you who found this package without caring about its specific features should do; download the PKGBUILD and use it locally. You won't get updates to the build script, but it should continue to work for a long while, on its own. That was always a feature :) download Firefox Nightly directly from Mozilla and install user-local, and always use the auto-update functionality. You'll keep most of the advantages but it's more work and perhaps more brittle. Keep on eye on the "unofficial user repositories" wiki page. It's been a good ride, I wish you all good times. ``` I think it's safe to say no one will miss it. FWIW, Felix, hosting a custom repo that packages firefox-nightly itself seems like the right way to go. Personally I'd use the source builds from https://bbs.archlinux.org/viewtopic.php?id=117157 (heftig is also the maintainer of the stable package in [extra], so he probably does a good job. ;)) -- Eli Schwartz Bug Wrangler and Trusted User
On 05/10/2018 05:36 PM, Félix Saparelli wrote:
How do I tell this request to, politely, go away?
always-nightly has a number of differences to -nightly, not just the "always" part of it. But if this *is* the main point of difference, then I contest the claim that -nightly has "pretty good tracking". As a blatant example, right now it is on 62.0a1.201805*09*, while always-nightly downloads 62.0a1.201805*10*/./
I'm curious if you've actually read the PKGBUILD. Or tried using it. It uses pkgver to update properly, using the version string from https://download-installer.cdn.mozilla.net/pub/firefox/nightly/latest-mozill... so even though the PKGBUILD hasn't changed since you first commented, it currently downloads: ==> Starting pkgver()... ==> Updated version: firefox-nightly 62.0a1.20180510-1 (I should note this is wrong too, but that's because curl gets a different, older file than my browser does. Also PGP "bad signature".)
There is also the fact that firefox nightly updates *twice a day*, not once, so the -nightly package is /never/ accurately tracking the latest.
Okay? The technical solution to that seems to be to add more granularity to the version extracted from said text file. -- Eli Schwartz Bug Wrangler and Trusted User
Request #11354 has been accepted by Eschwartz [1]. [1] https://aur.archlinux.org/account/Eschwartz/
participants (4)
-
Eli Schwartz
-
Félix Saparelli
-
Morgan Adamiec
-
notify＠aur.archlinux.org