Re: master b82adee: Invalidate dir-locals-directory-cache when writing dir-local file

[Štěpán Němec]

On Mon, 19 Aug 2019 13:05:58 -0400
Noam Postavsky wrote:

> On Mon, 19 Aug 2019 at 05:52, Štěpán Němec <address@hidden> wrote:
>
>> >>> Oops; forgot the "Copyright-paperwork-exempt: yes", and I guess it's too
>> >>> late to amend after I've pushed?
>> >> Too late.
>> >
>> > Yup.  Hard to believe we're still stuck with this problem of "unfixable
>> > commit messages" after all these years.
>>
>> There's "git notes", but I guess you're aware of that.
>>
>> <https://git-scm.com/docs/git-notes>: "A typical use of notes is to
>> supplement a commit message without changing the commit itself."
>
> Although this only adds to the message, right? A few years ago,
> someone proposed 'git replace --edit' for this, but I found it creates
> a lot of refs which cause noticeable slowdown for all git
> operations[1]. Not sure if 'git notes' would also create many refs.
> Andreas Schwab mentioned something about a new git refs backend to
> scale better[2]. Did anything ever come of that?
>
> [1]: https://lists.gnu.org/archive/html/emacs-devel/2016-07/msg01240.html
> [2]: https://lists.gnu.org/archive/html/emacs-devel/2016-07/msg01185.html

Ah, I knew the git notes suggestion seemed too obvious to be useful...
It seems good for adding a pseudo-header to an occasional commit after
the fact (without changing the commit object) such as here, but I'd
expect you hit the same scalability issues as demonstrated in your tests
with git replace, if used in bulk on a repo of this size. (I have no
data to back that up, though, neither can I comment on your second
question.)

-- 
Štěpán