Status of Transport work

[John A Meinel]

John A Meinel wrote:
> Hey Martin-
> I just wanted to check what the status was for my Transport branch. I'm
> pretty sure you've mentioned that you were going to merge it a couple of
> times, but obviously I haven't seen the patches show up yet.
>
> Is there some conflict preventing you from pulling them in, or something
> else that you are waiting for?
>
> Obviously I would like to see the SftpTransport updated to use paramiko,
> but I think the basic structure is correct, and we can update it from there.
>
> I would just like to see some commands like "bzr push" implemented, and
> I think the Transport layer would make it possible, since it allows
> writable remote access.
>
> If you have qualms for other reasons, that's fine. I'm not even sure if
> it would merge cleanly with the current sources (last I tried the merge
> operator didn't work with missing revisions in the history).
>
> Also, the changeset plugin in "bzr.dev/bzrlib/plugins/changeset" is
> *very* out-of-date.
>
> John
> =:->

In an off-list email you told me that all the tests didn't pass with the
latest version. The problem was actually only in the last couple of
patches (961..). I started working on implementing pipelined http
support. I had a little bit of success in time trials (ie the pipelining
helped), but I found out that the python HttpServer doesn't seem to
actually support pipelining (though it returns headers that made me
think it did).
Anyway, my get_multi() fails if you connect to a server that doesn't
support pipelining, and the changes were a little invasive (I made some
pretty strong changes to effbot.org in an attempt to have it re-use
connections rather than re-connect for each file).

I don't know if you want to investigate it. At present I don't have a
lot of time to be figuring out advanced HTTP protocols (handling
backwards compatibility, etc).

So for now, just merge up to revno 960. All the tests pass there. If you
want to work on Http pipelining, you can look into my other changes, but
I don't know if it is the route you want to go.

(There also seems to be a small bug that if the Http test doesn't
succeed it leaves the extra spawned process around, which then tries to
run the rest of the tests. Which is why you see a lot of the "directory
already exists" failures).

John
=:->

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 253 bytes
Desc: OpenPGP digital signature
Url : https://lists.ubuntu.com/archives/bazaar/attachments/20050817/8349ada9/attachment.pgp