Re: [TDF infra] Announcing Gitiles VCS browser (gitweb replacement) and https:// anon git URIs

On 10/31/2018 9:15 AM, Guilhem Moulin wrote:
> On Tue, 30 Oct 2018 at 16:18:36 +0000, Luke Benes wrote:
> > The date gitiles displays in the log is the author date, rather than
> > the much more useful commit date.
>
> Quoting my own <20181023142926.GB21836@localhost.localdomain>:
>
> | It's shows the authored date, not the commit date.  We can easily tweak
> | the template, but IMHO it's more consistent to show the authored date
> | given that the name that's beside it is the author's not the committer.
> | Gitweb, on the other hand, shows the *commit* date along with the
> | *author* name.
>
> > When trying to track down regressions, you care about the order of
> > commits, not when they were authored.
>
> I'm confused, neither commit date nor author date give any guaranty
> regarding the order of commits.  AFAICT the only thing that helps here
> is to follow the Merkel tree (child → ancestor) back to the root.
> Unlike ordering by date, performing a topological order of the tree
> (e.g., following the output from git-log(1)) is reliable, as the
> relations are cryptographically secured.

Well - of course, there's no guarantees; but given the workflow, when we 
have the author date recorded when the first version of commit was 
created, which might be months before the commit gets into the final 
shape (which can be drastically different from the first version), the 
"author date" has ~no value, while commit date has at least some, and is 
actually helpful.

Thus, given that

> I don't care which of author or committer is shown in
> the log view.  I believe it should be both committer name + date, or
> both author name + date for consistency, but if QA wants to mix the two
> then so be it.

please do the change :-)

-- 
Best regards,
Mike Kaganski