Troy,
Just letting you know that I'm reviewing this and it seems about ready to go.
So far, the main issue I've spotted is that the logic (functions) for detecting
and replacing "^/" belongs in the command-line client and not in the "path"
library.
- Julian
Troy Curtis Jr wrote:
>>From my original patch (http://svn.haxx.se/dev/archive-2007-11/0708.shtml)
> through discussions about a better implementation
> (http://svn.haxx.se/dev/archive-2007-11/1113.shtml and
> http://svn.haxx.se/dev/archive-2007-12/0047.shtml) I have returned with a
> better relative url implementation for the svn client, at least so I believe.
>
> In this patch I implement the following logic for parsing the svn CLI
> client args:
> - If there is a relative url in the target list
> - Then use the repository root url from the other targets to resolve the
> relative url.
> - If there are no other targets, use the root url of the current directory.
> - If the current directory is not a working copy, issue an error.
>
> This seems to get most of the use cases right. It does leave multi-repository
> urls out, but they are much less frequent and we couldn't come up with a
> simple (to understand) method of resolving them.
>
> This patch contains what is likely a ridiculous amount of svn tests, basically
> one for every command of interest to relative urls. It might be too much
> because ultimately they all run the same function. However, I had them from
> my previous attempt (in which each svn subcommand had it's own implementation)
> and I figured they could be useful in making sure checks and translations
> don't get made before the call to svn_cl__args_to_target_array() that break
> relative url support.
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe_at_subversion.tigris.org
For additional commands, e-mail: dev-help_at_subversion.tigris.org
Received on 2008-02-20 19:09:02 CET