[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: Why the split for rcvs?
Re: Why the split for rcvs?
Sun, 25 Mar 2001 03:18:20 +0800
Dear CVS developers,
I am watching the thread of "split for rcvs" and "locking and other patches"
and I must say that I am quite shocked.
Why is that the requirements for sending the patches are getting more hard
The conditions to accept the patches are quite too difficult to comply with
and simply designed to discourage the people from sending those ("I'm number
1 so why try harder" attitude?).
As a developer I would not expect that the users of my software would send
me a complete, fully tested and documented patch or bug fix. I don't even
want it - I know my software much better and I will certainly modify the
patch before applying. I will write the documentation myself, based on the
plain email description and my understanding of the system, I will make
corrections to the sent code etc. The effect will be better than pushing the
poor guy to write it himself or to make him exercise his skills in the areas
he is not familiar with.
I would also apply patch to the branch first, so people can try it and test
as an "experimental" version. That allows the patch to get mature and gives
some "battlefield-test" - always a good thing to do. When the thing is
ready - merge it to the trunk.
Now, back to the particular case of Noel's patches - why don't he gets the
write access to the repo? He did show the commitment and surprising patience
with those patches.
Or, why his patches are not put on the branch so people can check out the
code and build themself?
BTW: CVS code itself seems to have a surprisingly small number of
I don't understand why he has to re-do the patches against latest
development version - surely you should put it on the branch that was crated
for each release and merge with the trunk, no?
For the contrast - here is how we do things go on Windows. The cut of
changes to the cvsnt server project(www.cvsnt.org):
23/11/00: Added 'cvs commit -c' to complement 'cvs edit -c'
18/07/00: Removed 'admin -l' at last.
19/06/00: Merged in 'cvs edit -c' patch, <snip>
As you can see, at least the "edit -c" patch was applied, people are using
it for quite a while now with good results. Please note that there is no
more admin locking - so CVS server on Windows NT is fully concurent, with
professional solution for locking by means of Noel's patch. It seems that
Noel is not even aware of that - and that is how it is supposed to be! He
submits the patch and developers are picking it up and applying, testing,
accepting or rejecting. It should not be ignored, and there should not be
the patches "floating around" making a users chase them everywhere. Patches
should be with the main code, that is where people shuld search and find
Dear CVS developers, please give Noel a chance and better CVS to us all...
----- Original Message -----
From: "Derek R. Price" <address@hidden>
To: "Noel L Yap" <address@hidden>
Cc: <address@hidden>; <address@hidden>
Sent: Saturday, March 24, 2001 07:56
Subject: Re: Why the split for rcvs?
> Noel L Yap wrote:
> > Note that these patches are against cvs-1.11 and I have very minimal
> > them.
> You'll get the best results if any _submitted_ patches are against the dev
version. The less
> work that has to be done to get the thing properly into the tree as usably
and maintainably as
> possible, the more likely it is that someone with access will do it...
> Derek Price CVS Solutions Architect (
> mailto:address@hidden CollabNet ( http://collab.net )
> Old heads as well as young may sometimes be charged with ignorance and
> presumption. The natural course of the human mind is certainly from
> to skepticism.
> - Thomas Jefferson to Caspar Wistar, 1807
> Info-cvs mailing list