Re: Locking functional spec comments

Julian Foad wrote:

> + The working copy must store representations not only of its
> + own locks but of all other locks that affect it.

I think "must" is a bit strong here. Actually I don't see it's useful at
all; that information is stale almost by definition, and would
complicate the update process. How? Well, the state of locks on files
can change _without_ a bump to the global revision number, so what is
today an empty update would still have to pull all lock info frrom the
server. For example:

    A> svn update
    A> svn lock foo
    B> svn update (notes that foo is locked elsewhere)
    A> svn unlock foo (changes her mind, no change in the repo)
    B> svn update

At this point, the server notices that the client's working copy is
already up-to-date. But in order to maintain the lock status in the WC,
the server would _still_ have to send info about all locks in the
subtree. Yuck.

Storing knowledge about other people's locks in the WC will gain you
nothing, because any lock-affected operation would _still_ have to make
a round trip to the server to update stale info.

-- Brane

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Fri Oct 15 01:22:41 2004