PyCurlTransport doesn't define the _remote_is_at_least_1_2 property

[Mark Lee]

I'm trying to branch from a smart HTTP server, where the client uses pycurl as the http transport client. The following error occurs while branching:

bzr: ERROR: exceptions.AttributeError: 'PyCurlTransport' object has no attribute '_remote_is_at_least_1_2'

Is there a particular reason why that attribute is not defined in PyCurlTransport? If not, I have a simple patch which allows the pycurl transport to retrieve HTTP smart server branches.

A log of the console output and the relevant part of ~/.bzr.log is attached.

[Robert Collins (lifeless)]

On Tue, 2008-04-22 at 21:59 +0000, Mark Lee wrote:
> Public bug reported:
>
> I'm trying to branch from a smart HTTP server, where the client uses
> pycurl as the http transport client. The following error occurs while
> branching:
>
> bzr: ERROR: exceptions.AttributeError: 'PyCurlTransport' object has no
> attribute '_remote_is_at_least_1_2'
>
> Is there a particular reason why that attribute is not defined in
> PyCurlTransport? If not, I have a simple patch which allows the pycurl
> transport to retrieve HTTP smart server branches.
>
> A log of the console output and the relevant part of ~/.bzr.log is
> attached.

This smells familiar; didn't we fix this 3-4 weeks ago?

-Rob
--
GPG key available at: <http://www.robertcollins.net/keys.txt>.

[Martin Pool (mbp)]

It seems odd to me that this is on the http transport not the Remote object...

[John A Meinel (jameinel)]

class HTTPTransport
 ...
  def get_smart_medium():
    return self

The problem is that both urllib and pycurl inherit from HTTPTransport, which claims itself as the smart medium. And it is the *medium* where this is defined.

Back a few weeks ago, I mentioned that the smart medium tests were incomplete, if we were adding a *required* member to mediums, but not testing that they were actually there. This is just another one we overlooked.

[czp (christian-zipfel)]

I am running into the problem too.

I have
czp@wichtel:~$ bzr version
Bazaar (bzr) 1.4rc2
  Python interpreter: /usr/bin/python 2.4.4
  Python standard library: /usr/lib/python2.4
  bzrlib: /usr/lib/python2.4/site-packages/bzr-1.4rc2-py2.4-linux-x86_64.egg/bzrlib
  Bazaar configuration: /home/czp/.bazaar
  Bazaar log file: /home/czp/.bzr.log

on the smart server with bzr+http.

A newley updated Win32 client:
Bazaar (bzr) 1.3.1
   Python interpreter: C:\Python25\python.exe 2.5.1.final.0

which ran a bzr checkout, bzr add and bzr commit fine. (The Win32 version 1.3 encountered the AttribError).

And a MacOSX client:

nineteen:MatheZirkel czp$ bzr version
Bazaar (bzr) 1.3.1
  Python interpreter: /opt/local/bin/python2.5 2.5.2.final.0

where a
nineteen:MatheZirkel czp$ bzr update
bzr: ERROR: exceptions.AttributeError: 'PyCurlTransport' object has no attribute '_remote_is_at_least_1_2'

crashes.

[Andrew Bennetts (spiv)]

This has been fixed in bzr.dev. The fix will be in the bzr 1.6 release.

[pva (pva)]

Guys, speaking about currently stable 1.5 branch what it's better to do? Should we apply patch Mark posted here or it's better to wait for 1.6 release? Thank you.

[Mark Lee (malept)]

Well, 1.6 is scheduled to be released on Monday <https://lists.ubuntu.com/archives/bazaar-announce/2008-August/000178.html>, and my fix wasn't really optimal. If you want a patch in the meantime, I would probably use the revision committed by Andrew Bennetts:

http://bazaar.launchpad.net/~bzr/bzr/bzr.1.6/revision/3442

[pva (pva)]

Thank you very much Mark. I'll wait till monday and if release will be delayed for some reasons I'll take this patch as we want to use your overlay. )