Whole tree up to date before committing

[John Arbash Meinel]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Matthew D. Fuller wrote:
> On Thu, Oct 22, 2009 at 11:11:37PM +0200 I heard the voice of
> Nicholas Allen, and lo! it spake thus:
>>>> Wouldn't it make sense for those developers to work in their own
>>>> branches and only merge that to trunk when their branch is
>>>> considered complete?
>>> No. Most of the commits are not large features, but small bug
>>> fixes and additions. 
>> This is not a problem if you do each bug fix and addition in a
>> separate branch of its own.
> 
> I'm on Óscar's side here.  We commit directly on trunk _all_ _the_
> _time_, because most of our changes are small stuff that happens in
> one commit.  Doing those in a separate branch and merging is totally
> pointless overhead, and a policy demanding that would be a great way
> to get myself lynched.
> 

Note that I've had many 1-line "trivial" changes that ended up having
ramifications that I didn't expect, and which the test suite caught.
Perhaps not as many recently as I used to. However, "changes are small
stuff that happens in one commit" is not always "perfectly stable", and
I'm quite thankful that we have a PQM keeping trunk passing all tests.

(So that we know when merging/updating from trunk that failing tests are
our fault, not someone else.)

The few times when PQM misses parts of the test suite (unicode at one
point, still some issues with win32), it does have an impact on development.

John
=:->

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (Cygwin)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/

iEYEARECAAYFAkrg2CQACgkQJdeBCYSNAAPxEACgoyJVebXwAx0KU0nf/mXoDvxy
YRUAnj0iJc+4Z0LJxIi4/XAZuq0KlIka
=L/yA
-----END PGP SIGNATURE-----