Re: [PATCH] kbuild: pacman-pkg: make "rc" releases adhere to pacman versioning scheme

[Viktor Jägersküpper]

On 5/14/26 15:19, Nathan Chancellor wrote:
> On Thu, 14 May 2026 01:17:29 +0200, Viktor Jägersküpper <viktor_jaegerskuepper@xxxxxxxxxx> wrote:
>> diff --git a/scripts/package/PKGBUILD b/scripts/package/PKGBUILD
>> index 452374d63c24..fe629074b4e8 100644
>> --- a/scripts/package/PKGBUILD
>> +++ b/scripts/package/PKGBUILD
>> @@ -10,7 +10,7 @@ for pkg in $_extrapackages; do
>>  	pkgname+=("${pkgbase}-${pkg}")
>>  done
>>  
>> -pkgver="${KERNELRELEASE//-/_}"
>> +pkgver="$(echo "${KERNELRELEASE}" | sed 's/-rc/rc/;s/-/_/g')"
> 
> Sashiko notes [1] (with a contrived example) that this could result in
> custom localversions being matched and adjusted. While I don't think it
> is that big of a deal given how specific the example is, it takes little
> effort to make the match more restrictive to make such a situation less
> likely:
> 
>   pkgver="$(echo "${KERNELRELEASE}" | sed 's/-\(rc[0-9]\+\)/\1/;s/-/_/g')"
> 
> [1]: https://sashiko.dev/#/patchset/20030

I agree this is much better. Actually I should have looked at the
script "mkdebian" for Debian packages which has this line:

upstream_version=$("${srctree}/scripts/setlocalversion" --no-local "${srctree}" | sed 's/-\(rc[1-9]\)/~\1/')

This is incorrect for the (hopefully rare) case that we have 10 or more
"rc" releases, but that can be fixed in another patch.

Best regards,
Viktor