Re: svn commit: r11750 - branches/locking/subversion/include

Okay, there's been more IRC discussion between me, kfogel, fitz,
ghudson, and sander.
Of course, no IRC decisions are ever "official", so I'm re-explaining
the newest plan here on the list.

I think, ironically, that we're actually preferring to go with the new
opaque type after all, expanding on branko's idea. :-) It's
"cleaner" than attaching a username to the fs_t, yet making fs
functions fetch lock-tokens via callback.

If folks have time, take a look at the latest svn_fs.h on the locking
branch.

Here's the log message from my last commit:

"Forget the lock-token callback stuff.

Make the user_t into a more generic 'access' context. Allow this
access_context to hold all temporary user data, such as the
authenticated username and lock-tokens. When an fs function requires
either form of authentication, it can surreptiously pull stuff out of
the access_context as needed."

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Fri Nov 5 22:57:21 2004