Greg Stein wrote:
> Roll back a change? Dude. They are TWO line changes. If somebody
> doesn't like it, then just apply a change to remove it. "Roll back" is
> for large changes. These are about as small a change as possible.
The const poisoning is easier to remove in a commit. Yes, some of them were two
lines in total, but others touched functions in multiple files, say for a
function pointer typedef.
> They are harder to review because you must review a dozen emails. Put
> those dozen changes into one revision and just go thru it once.
That's for you, I don't agree. I find it easier to review smaller commits with
logically distinct changes, just a quick scan of the email if it looks right and
I can delete it. You like larger commits.
One reason it happened this way is that I would add a const to a function
declaration and then follow the compiler warnings adding const's where
necessary. Often times, I would end up in a static function that modified the
dereferenced value, so I would have to back out the const. It was easier just
to do a commit to get a const change in that worked and then continue on to the
next function in the header, which possibly wouldn't work, then I would have to
back out all the changes in several files and I couldn't use revert.
Blair
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe_at_subversion.tigris.org
For additional commands, e-mail: dev-help_at_subversion.tigris.org
Received on 2008-09-09 18:38:50 CEST