Re: [RFC] QA of manually set svn:mergeinfo

On 11/7/07, Paul Burba <pburba@collab.net> wrote:
> In the case of improperly ordered ranges I think we can safely say there
> is only one correct interpretation ("/trunk:6,3" == "/trunk:3,6" no
> matter how much you look at it :-).
>
> With overlapping changes with the same inheritability there is only one
> correct meaning as well ("/trunk:5-6,6-7" == "/trunk:5-7").
>
> *But* in both cases the fact that a user specified something odd raises
> a red flag for me that maybe they meant something else entirely or
> simply don't understand what they are doing. I'm quite comfortable
> being ultra-strict when someone is using propset to set svn:mergeinfo.

That was exactly my point. Your first example above is perfect. Was
the user just sloppy and ordered the value wrong ... or is the ",3"
supposed to be ",13"? There is no way to know. The fact that the
format is not right is a clue that something might be wrong, so maybe
we should just fail it.

If we are going to fail it, the error message has to help. They could
have thousands of revisions listed. If they put two of them out of
order we have to be able to tell them that or users are going to hate
us.

Think of the difference between saying an XML file is invalid and
telling you the line that is invalid. We need to try to help if we are
going to fail the command.

-- 
Thanks
Mark Phippard
http://markphip.blogspot.com/
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org