Re: ChangeLogging commits to CVS

Daniel Stone writes:
 > On Tue, 2004-12-14 at 16:12 +0100, Egbert Eich wrote:
 > 
 > I agree, and I also agree that having to break your changes out when
 > you've been working on things in parallel is a pain (but you can largely
 > sort this with -- write ChangeLog, 'cvs ci ChangeLog lib/Xpm/Imakefile';
 > write ChangeLog, 'cvs ci ChangeLog
 > programs/Xserver/hw/xfree86/driver/ati/radeon_driver.c', ...), but the
 > time it saves when working out what's been committed when is
 > immesurable, especially when you have to dig through tens or hundreds of
 > these huge, massively-unrelated changes.  Getting into the habit of
 > code-test-commit is probably also good, so we don't all have huge
 > bunches of code stewing locally, just ready to cause conflicts when we
 > all attempt to commit to about half the tree. :)

Often times there is a reason why I have so many different projects
floating around in parallel. One of course is that I have to work
on fixing many things in parallel. 
Often times there are other reasons: I may want to test things locally 
for a while before I dump it onto the public.
Also our policy to close down the entire tree for commits during the 
release process adds to this.
I tend to group small fixes together as the round trip times 
apply patch; write changelog; commit; for one-liners are just too 
time consuming.

I will try to improve but I can't guarantee that I will never ever
get carried away and commit two or three unrelated one line patches
in one batch ;-)

Cheers,
	Egbert.