I'm not sure I actually *like* asking people to consistently split their
commits up, though.  It's very convenient to make a variety of changes
in one commit, at our current size this doesn't feel like a major
problem, and it's certainly a widespread practice.  But perhaps it's
worth thinking about.

I'm in agreement with Ted - (plus, I like the term "omnibus commits" :)) - when you check in a massive change that breaks something, that is only asking for trouble.

This is why I advocate the frequent-checkpointing style of committing - each incremental change that you make, that leads to a code that still works, check it in. This holds even if you're going to check in the same file twice in one day.

For instance, when I was doing the sections, I made at least 2 passes at cleaning up and reorganizing the Table/Grid code, with no change to functionality.. I committed each time before committing any actual section work, even though I had section work in progress in my own tree. That way if I have broken something, I can figure out if it was failed refactoring or new code.

The double bonus is, if you've checkpointed just 5 hours ago and you realize that you've mostly just written garbage since then, at least you can throw your code out and go back to where you were 5 hours ago, rather than losing days of working off in your own tree.

Alec

Sincerely,
Jeffrey
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _

Open Source Applications Foundation "Dev" mailing list
http://lists.osafoundation.org/mailman/listinfo/dev

_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _

Open Source Applications Foundation "Dev" mailing list
http://lists.osafoundation.org/mailman/listinfo/dev

Reply via email to