Re: 4.3.0.1 Possible Release and Keeping XFree86 CVS Up to Date

[Brian Ford <ford at vss dot fsi dot com>]

On Wed, 30 Jul 2003, Alexander Gottwald wrote:

>On Tue, 29 Jul 2003, Harold L Hunt II wrote:
>
>> As for keeping the XFree86 CVS tree up to date, am I right that I
>> should be sending patches against XFree86 CVS xf-4_3-branch, instead
>> of the xf-4_3_0_1 snapshot?  Could I do this with:
>>
>> cd xc/programs/Xserver/hw
>> cvs diff -U3 -r xf-4_3-branch xwin > xwin.diff
>
-U3 is the same as just -u.  Cygwin, and most other projects (I don't know
about XFree86 for sure), usually also prefer -p to print C function
names in the diff header.  So:

cvs diff -up -r xf-4_3-branch xwin > xwin.diff

But, if your sources are already at the head of the branch (like they
should be to do this), the -r xf-4_3-branch is redundant.  If your
sources are on the branch, but out of date, this command would include
newer changes that have already been committed in with yours.  Since that is
probably not what you or they want, here too it is better to leave off
the -r xf-4_3-branch.  If your sources are not on the branch at all, this
is mostly just nonsense.

All that said, if you are the only one modifying xwin, it probably just
doesn't matter.

>the xf-4_3_0_1 snapshot is part of the xf-4_3 branch. So if you checkout
>the snapshot, you have the latest from the xf-4_3 branch. A simple cvs
>diff will create a diff against the xf-4_3 branch.
>
I disagree.  The snapshot is a point in time on the branch.  It may not
be, and probably isn't the latest.  A simple cvs diff will create a diff
against the snapshot, not the branch.

I would suggest just keeping your sources on the head of the branch.  The
only advantage to using a snapshot tag is reproduceability.  I don't think
snapshots have any sort of stability implied.

>> The main question here is can I make diffs against the specified branch
>> rather than the branch that I have locally?  I assume the answer is
>>yes, but some confirmation from others before I waste time looking into
>>it would be helpful.
>
Yes you can, but the result may be difficult to interpret.  As I said
before, you might get lucky just because you are the only one making
changes.  This is probably why they often have to hand apply your
patches, though.

>I think cvs diff -r tagname will do this. But it also includes all
>changes made in both branches.
>
I agree.  If you are on branch x, and you cvs diff with branch y, you
obviously get their difference.  If you apply that patch to branch y,
branch y ends up identical to branch x.  That might revert valid changes
on branch y, and thus is not a good idea.

I don't know if I have made things clearer, or just added to the
confusion.  I am pretty well versed in cvs, though.  So I can attempt more
clarification if needed.

-- 
Brian Ford
Senior Realtime Software Engineer
VITAL - Visual Simulation Systems
FlightSafety International
Phone: 314-551-8460
Fax:   314-551-8444