On Jan 13, 2005, at 6:30 PM, Max Bowsher wrote:
>
> At this point, I feel I have to defer to someone more knowledgable
> about the innards of mod_dav_svn regarding whether this feature's
> balance of increased complexity to usefulness is worthwhile overall.
>
That's me -- I'm one of the project's mod_dav_svn experts, and wrote
the autoversioning feature. And I'm here to rain on the parade. :-(
I really don't like these suggestions; the autoversioning feature was
written so that an subversion server could be minimally interoperable
with vanilla DAV clients. But both of these suggestions (the
content-type, and the log: header) are straddling this weird middle
ground. It's not pure DAV, but it's not DeltaV either. It's like "DAV
with extra custom stuff."
I don't think we should be in the business of creating this weird
middle zone. If you're writing an application to access a repository
via apache, you have three solid options:
* use the svn library APIs from C, perl, python, whatever.
This gives you maximum functionality and total control.
* use pure WebDAV commands (after turning on Autoversioning).
You get minimal features, but the command set is very tiny
and easy to use.
* use WebDAV/deltaV commands.
All features available, though the command set is larger
and more complex.
It's just way too messy to start developing a bunch of strange, custom
tweaks to pure WebDAV. It's not something we should take on,
support-wise... certainly not when there are already three valid tracks
to use.
I'd be glad to help anyone pursue any of these tracks.
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Fri Jan 14 01:49:49 2005