bzrlib.tests.blackbox.test_init.TestSFTPInit fails when run alone

[Robert Collins]

On Wed, 2007-07-18 at 21:18 +1000, Andrew Bennetts wrote:
> When running just the blackbox tests I see the following failure:
> 
> ----
> $ ./bzr --no-plugins selftest bzrlib.tests.blackbox.test_init.TestSFTPInit
> [...]
> 
> FAIL: blackbox.test_init.TestSFTPInit.test_init
>     not equal:
> a = ''
> b = "/var/lib/python-support/python2.5/paramiko/message.py:226: DeprecationWarning: integer argument expected, got float\n  self.packet.write(struct.pack('>I', n))\n"
> ----
> 
> This doesn't happen in the full run, because the deprecation warning is only
> shown once and gets triggered earlier when it is harmless.
...
> Also, it occurs to me that arguably deprecation warnings should not be shown to
> users at all.  Users can't really do anything sensible about API warnings except
> guess that upgrading their libraries *might* fix the problem.  It's developers
> (and packagers) that ought to be informed, not end users.  Perhaps distributions
> like Ubuntu should be suppressing DeprecationWarnings by default in released
> versions, and only display them by default for development versions?

This turns up with plugins as well, and was one of the sources of
complaints that prompted the strict policy some plugins adopted for
version-checking against bzr.

I propose:
 selftest runs have warnings enabled (and -Werror for pqm).
 bzr command line invocations have warnings/deprecations disabled by
default.

This test will still break, but it will not affect users.

-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/20070720/9aa43c45/attachment-0001.pgp