Greg Stein wrote:
> On Tue, Mar 30, 2010 at 12:36, C. Michael Pilato <cmpilato_at_collab.net> wrote:
>> Steve Simon wrote:
>>> Hi,
>>>
>>> Many appologies if this is not the correct place to ask about
>>> svn's dav support, if so please rediect me.
>>>
>>> AS per the subject, SVN's WebDAV server does not seem to
>>> be honouring partial HTTP GETs even though it appears to
>>> offer the facility with
>>>
>>> Accept-Range: bytes
>> [...]
>>
>>> Would you expect svn to support partial gets? Is SVN likely to support
>>> partial GETs in the future?
>> I wouldn't expect that support to exist today because Subversion's clients
>> don't require it (and that's really the driver for most of our server-side
>> features). Is it likely to grow that support in the future? Only if
>> someone writes the code. We've no reason to reject that enhancement, but
>> (again, because Subversion clients don't make use of the functionality) also
>> no compelling reason to prioritize it highly.
>
> Apache itself should be handling the ranged-GET operation. IOW, we
> shouldn't have to do anything. So maybe we broke Apache's range
> handling.
Apache doesn't pass along the range information to the content provider?
That seems rather inefficient, requiring that providers deliver the full
contents just so Apache can ignore some/most of them. Weird, but ultimately
off-topic here.
Glancing through the Apache sources, I'm guessing that byterange_filter.c is
the relevant code here. That filter has this comment:
/* Don't attempt to do byte range work if this brigade doesn't
* contain an EOS, or if any of the buckets has an unknown length;
* this avoids the cases where it is expensive to perform
* byteranging (i.e. may require arbitrary amounts of memory). */
I wonder if perhaps mod_dav_svn is creating naughty buckets?
--
C. Michael Pilato <cmpilato_at_collab.net>
CollabNet <> www.collab.net <> Distributed Development On Demand
Received on 2010-03-30 21:14:27 CEST