[darcs-users] Automated regression testing with each push

[Jason Dagit]

Alec Berryman <alec at thened.net> writes:

> Jason Dagit on 2005-05-02 07:39:56 -0700:
>
>> Thomas Zander <zander at kde.org> writes:
>> 
>> > ps. Darcs.net shows the manual which explains how to set this up.
>> 
>> I'd love to read documentation on this, but I've been unable to turn
>> up any.  I've tried google, and searching the wiki as well as the
>> table of contents for the manual.  If it's there, it's not obvious
>> to me how to find it.
>
> It's not particularly obvious, and Google and the wiki weren't
> particularly helpful for me, but it takes less than a minute of
> searching through the manual to find.

Ah, I had checked it by table of contents, for some reason I expected
it to have its own section.  I can imagine people who are looking at
the manual to find darcs features being impressed by this feature.
Perhaps it should be easier to discover...  Anway, thank you.

Now, I have few questions about what is written there:

  The ``_darcs/prefs/prefs'' holds the only preferences information
  that can propagate between repositories by pushes and pulls, and the
  only way this happens is when the setprefs command is used.

I don't fully grasp what that means, and it has been a source of
confusion for me in the past as well.  This is what I get out of the
above, and reading parts of the manual:

1) Manually editing _darcs/prefs/prefs only changes the local
   repository, and this is not really the standard way of making
   changes.

2) darcs setpref test "echo I am not really testing anything."  sets a
   test, and also causes this change to be propagated (right, now, at
   this minute).  Or is it that you now do darcs record, and it will
   be prpagated like any other patch?  Or the darcs record is not
   needed?

3) How does the test communicate success/failure to darcs?  What sort
   of information (if any) does darcs give to the script?

4) What are the steps taken to perform the test?  A copy is made in
   $TMPDIR and the test is run there?  Is this going to be a drain on
   the system resources? (ie does it just do a quick copy and test, or
   does it do a darcs get?)

I ask about #3, because our current commit script email the project
members notifying them who did an update (by username) and which files
were modified.   And #4 may or may not be important.

Thanks,
Jason