[PATCH] Display warning when flagging VCS packages

Eli Schwartz eschwartz at archlinux.org
Sun May 26 02:14:17 UTC 2019

On 5/25/19 6:52 PM, Lukas Fleischer wrote:
> On Sat, 25 May 2019 at 18:33:55, Bruno Pagani wrote:
>> Well I think it should instead be used for such bugs in the case of VCS
>> packages, because those are the only cases where they can be OOD, and in
>> contrary to normal packages those are valid OOD reasons. And that is the
>> case for the linked package in the FS ticket. But I acknowledge this is
>> not what we say currently, though I would use the opportunity of that
>> addition to change the guidelines regarding this. And then I’m in favour
>> of saying so in the message:
>> “This is a VCS package. Please do not flag it out-of-date if the package
>> version in the AUR does not match the most recent commit. Flagging this
>> package should only be done if the sources moved or changes in the
>> PKGBUILD are required because of recent upstream changes.”
> That message sounds good to me.

Me too!

>>>>> +/**
>>>>> + * Determine whether a package base is (or contains a) VCS package
>>>>> + *
>>>>> + * @param int $base_id The ID of the package base
>>>>> + *
>>>>> + * @return bool True if the package base is/contains a VCS package
>>>>> + */
>>>>> +function pkgbase_is_vcs($base_id) {
>>>>> +     $suffixes = array("-cvs", "-svn", "-git", "-hg", "-bzr", "-darcs");
>>>> What about false positives and missing items like -nightly’s? I think it
>>>> would be a good time to implement FS#56602, auto-seed the value
>>>> depending on your above list and let maintainers override this.
>>> Yes, there are false positives and false negatives. That is why we only
>>> display a warning and do not automatically disable the feature for VCS
>>> packages. Read the comments in FS#62733 for details.
>> All I’ve read was the same thing as before regarding the impossibility
>> to correctly detect all VCS packages and just them, but I did not see
>> why manual override wouldn’t be an option. ;) Regarding false positives,
>> without override possibility they will be misleading to users, so I
>> don’t agree on “it’s OK because we are not plainly disabling the
>> feature”. Also for me the strongest reason to not disable the feature
>> for VCS packages is rather because it is still useful even for those, as
>> stated by Eli. :)
> We could tune the message and say "This seems to be a VCS package."
> I would prefer to keep this very simple. That message is just for
> convenience and not really an essential part of the AUR.

Agreed on all points. The revised patch seems good to me.

Eli Schwartz
Bug Wrangler and Trusted User

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 1601 bytes
Desc: OpenPGP digital signature
URL: <https://lists.archlinux.org/pipermail/aur-dev/attachments/20190525/e1905788/attachment.sig>

More information about the aur-dev mailing list