Justin Erenkrantz <jerenkrantz@apache.org> writes:
> Adding a switch at run-time is code that we would never want to use
> permanently. Adding a temporary flag irks me somewhat.
>
> Furthermore, I believe doing this sort of experimentation would
> discourage long-term experimentation with Sander's diff library. If
> you select the diff library at configure-time, you don't have to
> actively remember to pass the flag. What we're looking for is
> long-term experimentation from people who would realize a broken diff
> when they see it.
>
> Passing an --internal-diff and telling everyone to sometimes use it
> isn't going to help. You actually have to use it all the time - we've
> got the common cases already down (via make test). Adding a configure
> flag and telling those people who aren't afraid to start eating
> Sander's delicious dog food would shake out the bugs a lot quicker.
>
> This is tending towards a bikeshed though. I'm just stating that I
> believe a run-time flag isn't going to advance the cause. -- justin
The main motivation for a run-time flag was to make it wasy to script
up some automated tests that compare expected merge outputs. (Note
that if it's a ~/.subversion/config option as well, then people don't
need to remember to use the flag each time).
Hmm, but your point about long-term use from people who would
recognize a problem is good too. A build-time flag is fine IMHO.
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Thu Feb 6 07:20:36 2003