Re: Process control for Subversion?

On Sun, 25 Jan 2009, Stephen Gross wrote:

> * Given a change request, initiate work on a codeline for that
> change request.

  my $0.02, for what it's worth. unless it's a trivial feature
request, start a new branch. and work on that branch until you're
good and done, and it's tested and you're happy with it, and you're
ready to merge back with the trunk.

  one of the things that really gripes my wagger is seeing numerous
commits in the log, most of them of the type, "oops, forgot X" or
"oops, fixing trivial bug in Y". that just clutters up the log with
lots of dicky little commits that breaks up the flow when you're
trying to read the history. and it makes reverting that feature a
real PITA since you have to revert all of those dicky little commits.

  use branches. branches are your friend.

rday

--
p.s.  i've even gone so far as to record all of the dicky little
commits that correspond to a single feature, revert all of those
commits in one operation, then check them all back in in a single
operation, just so i can have one commit for that feature.
is that anal retentive or what?
========================================================================
Robert P. J. Day
Linux Consulting, Training and Annoying Kernel Pedantry:
    Have classroom, will lecture.
http://crashcourse.ca                          Waterloo, Ontario, CANADA
========================================================================
------------------------------------------------------
http://subversion.tigris.org/ds/viewMessage.do?dsForumId=1065&dsMessageId=1060154
To unsubscribe from this discussion, e-mail: [users-unsubscribe_at_subversion.tigris.org].