bug#15310: 24.3.50; [ELPA] Better commit notifications for upstream authors

bug#15310: 24.3.50; [ELPA] Better commit notifications for upstream authors

[Dmitry Gutov]

Currently, when there are some changes made to a package that has an
upstream, an email gets sent for each group of commits pushed, to
upstream author (or authors, I'm assuming).

When the notification does come, getting the required changes out is not
as easy as it could be.

The commits are all sent together (one notification email per push,
AFAICS), as a plain-text email without attachments.

I have to identify the ones that originated from ELPA, and weren't
synced from upstream, or just sync commits, for each of them copy the
diff to a file manually, and separately copy the message and the author
when making the commit.

Since all paths the such diffs contain "a/packages/<package-name>", I
also have to run a replace command before applying the diff.

Is there a better way to work with these emails?

If no, it would be better if, for example, the script preparing these
emails identified commits not synced from upstream (maybe by file paths
inside the diffs?), fixed the paths, and attached these commits in patch
format that's accepted by 'git am' and already contains the commit
message and author information.

bug#15310: 24.3.50; [ELPA] Better commit notifications for upstream authors

[Stefan Monnier]

> The commits are all sent together (one notification email per push,
> AFAICS), as a plain-text email without attachments.

Indeed.  That's basically what Savannah's email notification gives us
(we can somewhat tweak it, the current config can be seen in
http://git.savannah.gnu.org/r/emacs/elpa.git/config, although we do not
have direct write access to it).

> Is there a better way to work with these emails?

"patch -p2" will strip off the "a/packages/" prefix.

> If no, it would be better if, for example, the script preparing these
> emails identified commits not synced from upstream (maybe by file paths
> inside the diffs?), fixed the paths, and attached these commits in patch
> format that's accepted by 'git am' and already contains the commit
> message and author information.

There's a copy of the script I use in elpa/admin/forward-diffs.py.
Improvements are welcome.

This said, I try to "push" the syncs separately from local changes, so
that an email should either contain a "sync" that you can completely
disregard, or only local changes.  This should be even more true once
I get a new cron job that does the syncs semi-automatically
(i.e. automatically when there's no conflict).  In that case you should
be able to pass the whole email to "patch -p2".


        Stefan

bug#15310: 24.3.50; [ELPA] Better commit notifications for upstream authors

[Lars Ingebrigtsen]

Stefan Monnier <monnier@IRO.UMontreal.CA> writes:

>> The commits are all sent together (one notification email per push,
>> AFAICS), as a plain-text email without attachments.
>
> Indeed.  That's basically what Savannah's email notification gives us
> (we can somewhat tweak it, the current config can be seen in
> http://git.savannah.gnu.org/r/emacs/elpa.git/config, although we do not
> have direct write access to it).

This was seven years ago -- is this still an issue?  (I haven't looked
at any upstream notification emails.)

-- 
(domestic pets only, the antidote for overdose, milk.)
   bloggy blog: http://lars.ingebrigtsen.no

bug#15310: 24.3.50; [ELPA] Better commit notifications for upstream authors

[Lars Ingebrigtsen]

Lars Ingebrigtsen <larsi@gnus.org> writes:

> Stefan Monnier <monnier@IRO.UMontreal.CA> writes:
>
>>> The commits are all sent together (one notification email per push,
>>> AFAICS), as a plain-text email without attachments.
>>
>> Indeed.  That's basically what Savannah's email notification gives us
>> (we can somewhat tweak it, the current config can be seen in
>> http://git.savannah.gnu.org/r/emacs/elpa.git/config, although we do not
>> have direct write access to it).
>
> This was seven years ago -- is this still an issue?  (I haven't looked
> at any upstream notification emails.)

There was no response, so I guess there'll be no further progress here,
and I'm closing this bug report.

-- 
(domestic pets only, the antidote for overdose, milk.)
   bloggy blog: http://lars.ingebrigtsen.no