On Wed, Feb 15, 2023, at 16:15, Guenter Roeck wrote:
> On Sat, Jan 28, 2023 at 01:29:04PM +0100, Ard Biesheuvel wrote:
>> Create a new status 'dead' which conveys that a subsystem is
>> unmaintained and scheduled for removal, and developers are free to
>> behave as if it's already gone. Also, automated build tests should
>> ignore such subsystems, or at least notify only those who are known to
>> have an interest in the subsystem in particular.
>>
>> Given that Itanium/IA64 has no maintainer, is no longer supported in
>> QEMU (for boot testing under emulation) and does not seem to have a user
>> base beyond a couple of machines used by distros to churn out packages,
>> let's mark it as dead. This shall mean that any treewide changes (such
>> as changes to the EFI subsystem, which I maintain) can be made even if
>> they might cause build or boot time regressions on IA64 machines. Also,
>> mark the port as scheduled for removal after the next LTS release.
>>
>
> Since this just came up, I very much prefer complete removal. I don't
> see the point of keeping dead code in the tree. That is still hidden
> maintenance effort.
>
> If this proliferates, we'll end up having to parse the MAINTAINERS file
> for code marked "Dead" to ensure that we don't accidentally send e-mails
> to the wrong people, or we risk getting complaints about sending reports
> for such code. That puts extra burden on maintainers of automated test
> beds, which I think is not really appropriate. If the code is dead,
> remove it, period.
>
> For my part, I'll drop my test bed support immediately after this patch
> made it in, following the guidance above.
I agree. While the idea of waiting for an LTS release makes sense
in general (and I did just that for the unused Arm board files), I
don't see how that would help with the timing here: The only remaining
distro with kernel updates is now Debian-ports, and the coming Bookwork
release will apparently use the 6.1-LTS kernel, but as I understand,
the mentioned (late-2023) LTS kernel will not be part either of the
following (mid-2025) Debian release or this one, so keeping it
for a year longer has all the extra cost without any real benefit.
Arnd
|