Code review comment for lp:~vila/bzr/move-test-servers

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

...
>>>> 2- bzrlib.transport.servers seems a bit longwinded
>>> I proposed transport.servers because:
>>> - They are specifically transport servers, not smart server servers, or
>>> git servers, or CVS servers; and I wouldn't really want to see a git
>>> server mixed in. I guess we could do servers.transports.*, but thats
>>> equally long winded, and IMO less clear.
>> Are they really? They are things that can be started, stopped, and
>> that have a URL. Their essentially connection to transports (as
>> opposed to the fact that they are only used for testing transports)
>> seems quite weak.
>
> They are all servers for VFS's, rather than servers for non-FS-like file
> systems.
>
> -Rob
>

Just to mention that the whole reason this was brought up at all, is
because the SftpServer implementation depended on something in
'bzrlib.tests' which caused a *runtime* dependency for sftp to require
having 'testtools' installed.

If these servers require 'bzrlib.tests' to be available, that certainly
seems like it belongs as 'test_servers'. (If you can't really use them
without loading the testing infrastructure.)

John
=:->

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (Cygwin)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/

iEYEARECAAYFAkt0GfgACgkQJdeBCYSNAAN34gCeNGiEfnLScEzhq1cBq0oKY13k
MfYAnifbuaB2Yxa9X84U4yMwIukLDgoA
=asMl
-----END PGP SIGNATURE-----