[Slackbuilds-users] [RFC] Adding features to .info format.

[Vladimir Nikishkin]

Willy Sudiarto Raharjo <willysr at slackbuilds.org> writes:

> [[PGP Signed Part:Undecided]]
>> There have been repeated discussions about two features that the current
>> .info file format is missing:
>> 1. aarch64 architecture. If in the past slarm64 was still an unofficial
>>     port, with -current it is official, and quite widely available, given
>>     the number of RPi machines available.
>> 2. urls of the form https://example.test/address/ and
>>     https://example.test/address/1.json , which are either not supported
>>     by wget or can be mixed with each other, if downloaded into the same
>>     directory, which is especially bad with Golang and Haskell builds,
>>     which have many package-components, called 1.json.
>> To address this issue, I propose a backward-compatible change to .info
>> files format.
>> 1. add DOWNLOAD_AARCH64 and DOWNLOAD_X86, a space-separated
>>     bash-string-list, identical in function to DOWNLOAD and
>>     DOWNLOAD_X86_64
>> 2. add DOWNLOAD_NAME, DOWNLOAD_X86_64_NAME, DOWNLOAD_AARCH64_NAME, and
>>     DOWNLOAD_X86_NAME, space-separated _optional_ strings, which, if
>>     present, specify what the results of download should be named. If
>>     they are absent, current logic is not changed.
>
> This is my personal opinion and does not reflect other admins:
>
> My take on aarch64 is NO official support in SBo, but we can take "ELSE IF" in the
> SlackBuild to pass needed flags to build if the maintainer is using aarch and have
> done some testing to make sure it builds fine on aarch64. I personally don't have
> RPi machines to test it and we don't have enough resources and time to handle 3
> architecture at the same time with over 8k scripts in SBo with only few active
> admins approving all the submissions that are coming in daily basis.
>
> I'm so grateful that Urchlay and Andrew has stepped up to help with the linter and
> semi-automatic CI engine in github. It really helped us to approve faster than
> before, but still requires some manual actions and time to make sure everything
> works as expected.
>
> i can't imagine how long would it take for the next repository to be ready if we
> have to test 8k+ scripts x 3 architectures using the number of resources we have.

This RFC is not trying to make AARCH64 support mandatory, but is merely
suggesting to allow such an entry in the info files. (I.e., allow
sbolint to not bail out.)

-- 
Your sincerely,
Vladimir Nikishkin (MiEr, lockywolf)
(Laptop)