Re: CEDET Merge

[David Engster <deng@randomsample.de>]

Edward John Steere writes:
> David Engster <deng@randomsample.de> writes:
>>> Edward John Steere writes:
>>>> As I said in my previous email I'll make a start on the tests.
>>>
>>> That's great, thanks! I'll push a branch with my progress soon.
>>
>> I just pushed my first try as scratch/last-cedet-merge.
>>
>> -David
>
> I've just pushed up a fairly rough attempt at merging the tests in
> scratch/merge-cedet-tests

Thank you.

> (it looks like tests were only ever merged to Emacs once.)

Yes.

> I took the following approach:
>
>  * I used git to format patches per test file.
>  * changed the destination of the file in each patch to match the Emacs
>    manual testing directory;
>  * fixed any commit messages which failed the commit hook;
>  * added a final commit to cleanup; which included the removal of
>    duplicated test files not moved by the patching process and the
>    removal of dependencies on language/project support which we're not
>    merging
>
> The consequence of this approach is that my branch will add 316 commits.
> Many of the commit messages aren't up to scratch (they pass the commit
> hook but don't match the requirements outlined in CONTRIBUTE).  I'm
> going to have to spend some time fixing the rest of the commits.  I
> wanted to ask whether we should consider squashing all 316 commits into
> a "cedet-merge" commit since the changes are going to be documented in
> the ChangeLog.

There seems to be a misunderstanding here: The ChangeLog is generated
from the commit log and not written separately anymore.

I don't think it makes sense to fix up all the commits messages. The
commits you've merged go back to the beginning of CEDET, and I don't
think there's any sense it writing proper ChangeLogs for them now. They
are only tests, after all. However, it is very good that we have the
history on your branch, because it makes it much easier to check if all
authors have signed papers (last I asked, authors of non-trivial tests
were also required to have papers signed with the FSF).

In my opinion, once we have decided which tests to keep and fixed them,
we should squash them and commit it as a new test suite.

> Finally; the tests don't run with the CEDET in Emacs.

That's totally expected. I'll take a look at them.

-David