RE: Build system info needed
From: Kopp David K <david.kopp_at_us.army.mil>
Date: 2003-10-29 17:51:29 CET
Shamim,
I am willing to reconsider.
Here is how I understand what Greg is talking about (As an aside, I have
This approach is to add code to svn_repos_open in repos.c in libsvn_repos
The resulting function would look something like this:
svn_error_t *
/* If we are supposed to make an external lock file, do it here. If we
/* Fetch a repository object initialized with a shared read/write
SVN_ERR (get_repos (repos_p, path,
return SVN_NO_ERROR;
My question then becomes, where is the lock file deleted? I didn't see any
Is this right?
David
-----Original Message-----
Dude,
Kopp David K said:
See Greg Hudson's post.
It's not a kludge. Having a pre/post condition for a function is not unheard
It's actually a valid idiom - design by contract.
Should you choose to go down that path, it's entirely feasible that we could
Or possibly externalize it someday into a python hook or what not.
And you avoid syncrhonization issues between the ra_file protocol and yours.
There are a myriad of issues avoided. Maybe it's worth reconsidering?
-- Shamim Islam BA BSReceived on Fri Oct 31 14:37:22 2003 |
This is an archived mail posted to the Subversion Dev mailing list.
This site is subject to the Apache Privacy Policy and the Apache Public Forum Archive Policy.