Re: Monday Patch Queue Review update (2023-04-03)

[Carlos O'Donell <carlos@redhat.com>]

On 4/4/23 15:09, Sergey Bugaev wrote:
> On Tue, Apr 4, 2023 at 8:48 PM Carlos O'Donell <carlos@redhat.com> wrote:
>> Your patches:
>>
>> (a) Made it to the mailing list.
>>
>> (b) Made it into Patchwork (which we use for patch tracking)
>>
>> (c) Were reviewed as part of the weekly patch queue review.
>>
>>     - We look over patches in the meeting to see if we can help
>>       move them forward.
>>
>> (d) Did not get assigned any specific reviewer to review them.
>>
>>     - This happens for any number of reasons e.g. lack of a person
>>       who feels qualified to review a subsystem or machine,
>>       lack of hardware to test, etc.
> 
> Thanks for the explanation!
> 
>> The outcome of the meeting was that we didn't find a way to help
>> move your patches forward, sorry for that.
> 
> Ah, well, based on my previous experience, we just have to patiently
> wait for Samuel to find some time to review the patches :)
> 
>> Your personal queue is here:
>>
>> https://patchwork.sourceware.org/project/glibc/list/?submitter=35358
>>
>> Please have look over the queue to see if some of those patches have
>> been committed or could have their state updated.
> 
> Cool -- but I see that Patchwork gets confused by my somewhat liberal
> use of patch series formatting:
> 
> * "[v2] hurd: Add expected abilist files for x86_64" supersedes
>   "[RFC,34/34] hurd: Add expected abilist files for x86_64", so the
>   latter should have State = Superseded, and the former State = RFC

RFC and Superseded markup is manual today.

Looking at libc-alpha I see the following nesting:

[RFC PATCH glibc 34/34] hurd: Add expected abilist files for x86_64   Sergey Bugaev
    [RFC PATCH glibc 34/34] hurd: Add expected abilist files for x86_64   Florian Weimer
        [PATCH v2] hurd: Add expected abilist files for x86_64   Sergey Bugaev
        [PATCH v2] hurd: Add expected abilist files for x86_64   Florian Weimer 

Patchwork considers the "[PATCH v2] hurd: ..." to be distinct patches.

I believe that if you post with the subject e.g. git format-patch --rfc -v2:

"[RFC PATCH v2 34/34] ..."

Then it may update the patch to v2, but I haven't confirmed that.

> * The same goes for "[v2] hurd: Implement sigreturn for x86_64" and
>   "[RFC,32/34] hurd: Implement sigreturn for x86_64"

Likewise.

> * Ditto for the "Alignment-respecting x86_64 trampoline.c"
>   mini-series, which collectively supersedes
>   "[RFC,18/34] hurd: Port trampoline.c to x86_64"

I've never seen mini-series formatting like this. Is it used somewhere else?

> (but it did grok that [PATCH v2 18.0/34] means a cover letter! unless
> that was done manually)

It does that manually for the 0th entry in a series.

> Is there anything I should do differently when sending a replacement
> for a patch (but not a v2 of the whole series) to make it easier for
> Patchwork to understand what's going on?

I suggest trying my comments above and we'll see if that works.

I do not suggest using a mini-series. If you have to do a mini-series I think it
would be better to post a v2 with a renumbered set of patches.

-- 
Cheers,
Carlos.