New test files in emacs-25 branch

New test files in emacs-25 branch

[Eli Zaretskii]

I guess new test files introduced on the emacs-25 branch will have to
be manually added to the corresponding directory on master, since
gitmerge.el won't be able to pull such trickery, right?

Re: New test files in emacs-25 branch

[Artur Malabarba]

2015-12-03 11:37 GMT+00:00 Eli Zaretskii <eliz@gnu.org>:
> I guess new test files introduced on the emacs-25 branch will have to
> be manually added to the corresponding directory on master, since
> gitmerge.el won't be able to pull such trickery, right?

Are test files part of the release? If not, we should just add any new
test files to master.

Re: New test files in emacs-25 branch

[Eli Zaretskii]

> Date: Thu, 3 Dec 2015 12:53:18 +0000
> From: Artur Malabarba <bruce.connor.am@gmail.com>
> Cc: emacs-devel <emacs-devel@gnu.org>
> 
> 2015-12-03 11:37 GMT+00:00 Eli Zaretskii <eliz@gnu.org>:
> > I guess new test files introduced on the emacs-25 branch will have to
> > be manually added to the corresponding directory on master, since
> > gitmerge.el won't be able to pull such trickery, right?
> 
> Are test files part of the release?

That's a separate issue.  They weren't until now.

> If not, we should just add any new test files to master.

If they are not on the branch, how can I test the branch's build?

Re: New test files in emacs-25 branch

[Phillip Lord]

Eli Zaretskii <eliz@gnu.org> writes:

> I guess new test files introduced on the emacs-25 branch will have to
> be manually added to the corresponding directory on master, since
> gitmerge.el won't be able to pull such trickery, right?


I guess this crossed my mind sometime after I offered to move all the
file locations. Probably it would have made sense to leave things till
after the emacs-25.

If this is going to be a major PITA (and it probably is), one solution
would be to create a "test/automated" directory on master. The make
files will already pick this up correctly and run any tests in it (on
master). These files can then be moved enmass after the emacs-25
release.

Is git is intelligent enough to work through a file move? Will new tests
added to existing test files in the automated directory of emacs-25
merge into the moved files on master? If not, I fear, new tests in
existing files is likely to be more of an issue. If so, then new files
in emacs-25 test/automated can be merged to master, and then moved to
the new directory structure any time on master, and new tests or changes
should follow.

Phil

Re: New test files in emacs-25 branch

[Eli Zaretskii]

> From: phillip.lord@russet.org.uk (Phillip Lord)
> Cc: <emacs-devel@gnu.org>
> Date: Thu, 03 Dec 2015 17:52:22 +0000
> 
> Eli Zaretskii <eliz@gnu.org> writes:
> 
> > I guess new test files introduced on the emacs-25 branch will have to
> > be manually added to the corresponding directory on master, since
> > gitmerge.el won't be able to pull such trickery, right?
> 
> I guess this crossed my mind sometime after I offered to move all the
> file locations. Probably it would have made sense to leave things till
> after the emacs-25.

It wouldn't have mattered.  Emacs 25.1 is not going to be the only
version delivered from the emacs-25 branch.

> If this is going to be a major PITA (and it probably is), one solution
> would be to create a "test/automated" directory on master.

It's not a PITA.  For a rare event such as this one, committing twice
is not a catastrophe.  I just wanted to be sure there are no better
ideas.

> Is git is intelligent enough to work through a file move? Will new tests
> added to existing test files in the automated directory of emacs-25
> merge into the moved files on master?

That should work, but I wasn't talking about this.  I was talking
about adding new files to emacs-25.

> If so, then new files in emacs-25 test/automated can be merged to
> master, and then moved to the new directory structure any time on
> master, and new tests or changes should follow.

That's error prone, so I'd rather we avoided it.

Re: New test files in emacs-25 branch

[Phillip Lord]

Eli Zaretskii <eliz@gnu.org> writes:
>> I guess this crossed my mind sometime after I offered to move all the
>> file locations. Probably it would have made sense to leave things till
>> after the emacs-25.
>
> It wouldn't have mattered.  Emacs 25.1 is not going to be the only
> version delivered from the emacs-25 branch.

Really? I assumed 25.2 was going to fork from current master.

>> Is git is intelligent enough to work through a file move? Will new tests
>> added to existing test files in the automated directory of emacs-25
>> merge into the moved files on master?
>
> That should work, but I wasn't talking about this.  I was talking
> about adding new files to emacs-25.

Sure, I understand that. I was thinking of the situation where a new
file is added to emacs-25, merged to master and then more tests are
added (i.e. the new files is subsequently updated). As you say, the
addition of a file into two places is not so bad. But it would be nice
if subsequent updates to the file worked.

>> If so, then new files in emacs-25 test/automated can be merged to
>> master, and then moved to the new directory structure any time on
>> master, and new tests or changes should follow.
>
> That's error prone, so I'd rather we avoided it.

I'm at the limit of my git knowledge, I fear.

Phil

Re: New test files in emacs-25 branch

[Eli Zaretskii]

> From: phillip.lord@russet.org.uk (Phillip Lord)
> Cc: <emacs-devel@gnu.org>
> Date: Thu, 03 Dec 2015 22:41:31 +0000
> 
> Eli Zaretskii <eliz@gnu.org> writes:
> >> I guess this crossed my mind sometime after I offered to move all the
> >> file locations. Probably it would have made sense to leave things till
> >> after the emacs-25.
> >
> > It wouldn't have mattered.  Emacs 25.1 is not going to be the only
> > version delivered from the emacs-25 branch.
> 
> Really? I assumed 25.2 was going to fork from current master.

It could be, but given our previous history, it rarely happens.  The
master branch gets many new features very quickly, so making a minor
release from it no longer makes sense.  That is the price we pay for
not freezing master when a release is being prepared.

> I was thinking of the situation where a new file is added to
> emacs-25, merged to master and then more tests are added (i.e. the
> new files is subsequently updated). As you say, the addition of a
> file into two places is not so bad. But it would be nice if
> subsequent updates to the file worked.

I don't expect them to work.