[MERGE] Reading pack-names triggers 3 reads

[Robert Collins]

On Tue, 2008-11-11 at 22:30 +1000, Martin Pool wrote:
> On Tue, Nov 11, 2008 at 7:59 PM, John Arbash Meinel
> <john at arbash-meinel.com> wrote:
> > 3) Use some sort of stat() call, and implement some sort of stat() for
> > HTTP. This still causes us to issue 2 round-trip requests in order to
> > read pack-names, which is bad. Though I suppose the code could be:
> 
> (As you may already know) we could use http HEAD (try eg "curl --head
> http://bazaar-vcs.org/bzr/.bzr/repository/pack-names").  But, of
> course you're right that this issues two round trips, and for small
> files it will be no faster than reading the whole thing.

A partial read (so a readv) of the first page will return the whole
length (always, and its free); its a layering/functionality issue in our
HTTP transport that we can't access that data.

-Rob
-- 
GPG key available at: <http://www.robertcollins.net/keys.txt>.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: This is a digitally signed message part
Url : https://lists.ubuntu.com/archives/bazaar/attachments/20081112/2d8cd0ad/attachment-0001.pgp