Re: In defense of Link (was Re: [b4] initial "b4 dig" to supplant Link: trailers)

["Vlastimil Babka (SUSE)" <vbabka@kernel.org>]

On 10/13/25 01:27, Jason Gunthorpe wrote:
> On Sat, Oct 11, 2025 at 11:33:41AM +0100, Mark Brown wrote:
>> On Sat, Oct 11, 2025 at 08:13:51AM +0200, Willy Tarreau wrote:
>> 
>> > Maybe what would be needed would be a "Lore" (or "Series" or "Discussion")
>> > trailer that automatically makes a lore link from the message-id, and
>> > which would explicitly indicate that there's nothing else there but the
>> > actual patch submission, contrary to Link that's for any extra info ?
>> > It would be added by "git am" when passed a specific option.
>> 
>> There was an attempt to do that with the https://patch.msgid.link
>> domain, that's basically the same idea just done with a domain instead
>> of the name of the tag.  Linus wasn't any happier with that idea
>> unfortunately.
> 
> FWIW I thought that was a great compromise and have been trying to
> follow the 'if it is mechanical information it is msgid.link' and 'if
> it is deliberately meaningful then it has a [1] footnote and a lore
> URL'
> 
> IMHO, I think a 'b4 dig' like tool is a great way to address Linus's
> use case of: 'show me all mailing list discussions about this commit
> and highlight what might be interesting'
> 
> I think including the msgid.link would make this job of b4 dig work
> better.
> 
> Like others in the thread I use the Links directly quite a bit, it
> works well and is convenient.
> 
> eg I posted a bisection found regression 6 months after the series
> was accepted and got a fix:
> 
> https://lore.kernel.org/lkml/20250917153209.GA2023406@nvidia.com/
> 
> Responding to the original series is the easiest way to get a solid
> list of interested parties.

That's all nice and everything Jason, but Linus has been very clear on this,
so it's just beating a dead horse now.

> Jason
>