Long ago, we decided (sensibly, imho) that Subversion would set a
policy of using UTF-8 for log messages. Technically, we didn't have
to do this, but it aids interoperability by preventing encoding
anarchy.
Yesterday, Jon Trowbridge pointed out that we really ought to do the
same for property values. We already do it for certain properties
whose values are interpreted by Subversion itself (svn:ignore,
svn:keywords, etc). Jon's point is, why not do other clients the same
favor we do ourselves? Textual data is useless if you don't know the
encoding, and if we're hoping for unexpected property usages to arise,
it'll help a lot to define a standard here.
One response, of course, is "You can't do that, because property
values are binary!" But no one seems to have needed a binary property
yet, and it seems much more likely than property values would be text.
Anyone can base64-encode to get binary values if they really need
them, but optimizing for textual data seems reasonable.
So the proposal is: let's declare all property values to be textual,
and do UTF-8/eol conversion. Then svn_prop_needs_translation() can go
away, and every caller will just do translation unconditionally. We
will recommend that those who need binary prop values base64 encode
them.
Any objections?
-Karl (& Mike, Ben, Gstein, Jon)
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Wed Dec 4 18:00:17 2002