On Sun, 12 Aug 2007, Jean-Marc Lasgouttes wrote:
Abdelrazak Younes <[EMAIL PROTECTED]> writes:
Lars, as always, I simply cannot understand you and vice-versa. So
please, work as you like and let me work as I like. I won't listen to
you in any case...
Hello there,
I am always impressed to see how you manage to speak past each other
:) But if I may intervene, I'd say that there is a fairly strong
consensus here in Bromarv that a branch is not done to be dumped on
the trunk.
My impression as well. I'd go further and say that committing things
without previous review[*] is _not_ appreciated. As I understand it (which
may well be wrong of course), patches allow others to review the changes
before committing. If a single patch is big, it's much more difficult to
look at it :-(.
If a change in a branch is so distributed/complicated/diffiult that it
can't be converted into several managable patches, something else is
needed. Perhaps some kind of overall documentation or design document
(complementing the commit/log messages)? That would at least help other
developers to review the changes inside the branch more effectively.
Has anyone other than you reviewed the branch?
Best regards from Bromarv
Christian
[*] According Andre', there's an exception in effect here at the meeting
where the particpants may commit at will... Since Andre' has broken the
build for me many times these days, I can see that he likes the expection
here, but also why it's a very good idea with reviews.
However, Lars' XML branch will not be merged into trunk as a single chunk,
but rather in minimal patches with #ifdef clauses surrounding new code to
minimize impact to other developers.
--
Christian Ridderström, +46-8-768 39 44 http://www.md.kth.se/~chr
