On Thu, Nov 6, 2008 at 3:11 PM, Blair Zajac <blair_at_orcaware.com> wrote:
> David Glasser wrote:
>>
>> On Thu, Nov 6, 2008 at 10:36 AM, Blair Zajac <blair_at_orcaware.com> wrote:
>>>
>>> Greg Hudson wrote:
>>>>
>>>> On Thu, 2008-11-06 at 09:28 -0800, Blair Zajac wrote:
>>>>>
>>>>> Where I'm coming from is treating the svn_fs.h API as a versioned
>>>>> filesystem, ignoring the version control aspects of svn, so making this
>>>>> available via HTTP could be useful and have interesting applications
>>>>> available in the future.
>>>>
>>>> Subversion's architecture has never included making the FS interface
>>>> available over the network. There's no reason someone can't make a
>>>> network binding to the FS layer (as you did), but that's not in any way,
>>>> shape, or form the job of the Subversion RA layer.
>>>
>>> There's two different things as I see it, correct me if I'm wrong.
>>> There's
>>> the RA layer in C which will use the protocol. This won't change and I'm
>>> not suggesting modifying it. I'm just saying, make the new protocol
>>> capable
>>> of making the FS interface available over the network. Other people can
>>> write their own code on top of the protocol separate from the RA layer.
>>
>> That sounds like a great third-party project, but I don't see why it
>> has any business being part of the already-overextended subversion
>> core.
>
> I'm not saying we write that other code, I'm just saying the protocol should
> support it.
Now *that* would be a complete and utter rewrite of the existing HTTP
protocol. Or at least a gigantic boatload of new request types. The
current HTTP protocol is "mostly" a mapping to the RA vtable; it
looks very little like svn_fs.h. I'm not signing up for such a task.
:-)
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe_at_subversion.tigris.org
For additional commands, e-mail: dev-help_at_subversion.tigris.org
Received on 2008-11-06 22:19:32 CET