On 1/23/2012 8:52 PM, Daniel Shahaf wrote:
Personally I don't see any obvious reason for deciding that svn property names can be, of all things, "any XML name" --- the fact our wire protocol uses XML need not ripple into our public API; ..
I'm getting sleeping and I'm heading to bed, but I just wanted to add that you make two separate points there, and on their own they are completely valid. If we were sitting down writing the Subversion Specification (I mean the hypothetical specification that in some perfect world was written before any implementations), I would be wholeheartedly agreeing with you on both counts. Very well put.
But since there are implementations already in the wild, I'm a bit biased as well as disgruntled over all this "erroneous" data the DAV+SVN client allowed me to create. Even if we ignore my case, I wonder if there might be other users who might become infuriated if we made DAV_SVN suddenly started rejecting their applications which had worked for years. So I'm wondering if expanding the client's idea of valid property names is the lesser of two evils.
2. That there be a public specification that rigorously defines what makes a valid Subversion property name to prevent contradictory implementation issues like this in the future... but +1 on deciding what a property name is and being consistent about it. (with the caveat of not breaking existing repositories, per r1235131)
Thanks! Good night, Garret
