Re: Patch review proces WAS: Patch to align virt and phys spaces on i386

Hi,

I hope you don't mind some comments from a otherwise inactive 
developer :-)

Am Donnerstag, 9. März 2006 14:06 schrieb Timothy Legge:
> Hi
> 
> I am certainly fine with this approach.  Is there any guidance on how
> this process will work?  Some questions I have are:
> 
> 1) Can we commit to branches of the cvs to work on code we would like to
> have in source control prior to submitting it?
> 2) How will the patches get approved?  Should it require comment from a
> few developers?  Which ones?
> 3) Should we also attach the patches to Sourceforge to better track
> whether they have been applied?

Too much work IMHO. Sending a mail with attachment is super quick, logging 
in to sf and go to the patch manager takes several times longer.

> 4) Should a third patches mailing list be created?

Nobody will read it. IMHO the two existing lists are fine. Usually all 
developers are also subscribed to the users list, but not every user 
wants to hear all that technical stuff.

> From what I understand, patch management is handled differently by many
> projects.  Some require the user to resubmit the patch until it is
> accepted or rejected (lost in email).  Others provide voting on a patch
> or feature.  I assume that lack of comment is not implied approval.
> Does someone need to pick up the "trivial" patches that are occasionally
> submitted and submit thim in accordance with the patch process?

We have made good experience in LyX with the following procedure:

- All patches are sent to the mailing list before committing
- If you think that it is controversal stuff, ask for comments. This will 
usually start a discussion and often end with a conclusion and a revised 
patch.
- If it is some obvious bugfix, say that you will commit it in a minute 
and do so
- If you think that the patch is uncontroversal, but are not really sure, 
say that you will commit it in a day or so if nobody complains before.

Before a release a freeze is announced and nobody is allowed to commit 
anything without an explicit OK from the boss. This can be really 
annoying, because you may have to send ping messages several times, but 
is obviously necessary to reach a stable state.
This procedure helps to keep the review process running with as little 
bueraucracy as possible.


Georg




-------------------------------------------------------
This SF.Net email is sponsored by xPML, a groundbreaking scripting language
that extends applications into web and mobile media. Attend the live webcast
and join the prime developer group breaking into this new coding territory!
http://sel.as-us.falkag.net/sel?cmd=lnk&kid0944&bid$1720&dat1642