Re: squid3 PRE1 ?

On Monday 21 April 2003 11.08, Guido Serassio wrote:

> OK, but the next patch to win32.cc will be huge because contains
> all the win32 specific service code.

The size of a given patch does not indicate how troublesome the patch
is.

The number of places modified, and the quality of these modifications
to the existing code is mainly what matters.

A huge patch which primarily adds code used in specific conditions is
a trivial patch to review. The reviewer then only has to consider
  - Is this feature really useful
  - Is the interactions with existing code well defined
  - And a quick assessment of the quality of the added code

A small patch which makes unobvious changes at different places in the
code is a very troublesome patch, and won't be accepted even in
DEVEL, and will be rejected with a request to make another patch
which makes the code more obvious, or at a minimum extremely well
documented (but it is preferred if code does not require
documentation to be understood).

A huge patch which changes an internal design to a better one mainly
requires the patch to be backed up with a description of why the
original design sucks and why the new one is better. These changes
are fairly open in DEVEL, but requires discussion in PRE and should
not be seen in STABLE unless required in order to fix important bugs.

Note: In terms of places modified less is not necessarily better than
more. If a patch needs to make modifications to existing code for
best overall quality then it should. Trying to limit the amount of
existign code modified at the expense of the quality of the added
code is not acceptable as a design goal.

Regards
Henrik
Received on Mon Apr 21 2003 - 06:30:18 MDT