xmlrpc: Bugzilla integration issue

[Jim Hague]

On Tuesday 28 Feb 2012 19:11:46 Oleg Smolsky wrote:
> On 2012-02-28 11:05, Jim Hague wrote:
> > On Tuesday 28 Feb 2012 17:18:18 Oleg Smolsky wrote:
> >> Hey there, I've just hit an issue with Bugzilla integration in
> >> hgext/bugzilla.py (xmlrpc method).
> >> [...]
> >> 
> >>   Could someone shed some light on this please? It almost seems that the
> >> 
> >> code chokes on the compressed HTTP payload...
> > 
> > Rats. You are quite right. I think that's exactly what is going wrong :-(
> 
> Thank you for the quick reply, Jim! I've just debugged this further and
> the fix is to do the following inside the extension:
> 
>          # Deal with compression.
>          # ToDo:
>          #    this needs to check the Content-Encoding header...
>          import StringIO
>          data = StringIO.StringIO(payload)
>          import gzip
>          gzipper = gzip.GzipFile(fileobj=data)
>          payload = gzipper.read()

Yes, that would work, though as you say I'd need to check the Content-Encoding 
header.

Since the API calls are not large, and I don't expect much data to be 
transferred, my feeling is that compressing the traffic isn't a big win here 
anyway. Oleg, can you try this tiny patch instead and let me know if it works 
for you?

diff -r 476a981fdf34 -r cd83a2e41c9a hgext/bugzilla.py
--- a/hgext/bugzilla.py Mon Jan 16 01:21:30 2012 -0600
+++ b/hgext/bugzilla.py Wed Feb 29 11:00:18 2012 +0000
@@ -497,6 +497,7 @@
 
     def request(self, host, handler, request_body, verbose=0):
         self.verbose = verbose
+        self.accept_gzip_encoding = False
 
         # issue XML-RPC request
         h = self.make_connection(host)
-- 
Jim Hague - jim.hague at acm.org          Never trust a computer you can't lift.