[TIP] Fwd: Bizarre results running Hudson job + CI

[Jesse Noller]

On Mon, Mar 22, 2010 at 11:12 AM, Olemis Lang <olemis at gmail.com> wrote:
> After a few days of frustration  :( , I'm forwarding the message to
> this list (TiP ;o) because probably somebody has a clue and can help
> me to find the underlying issue.
>
> Sorry for cross-posting
> :-/
>
> ---------- Forwarded message ----------
> From: Olemis Lang <olemis at gmail.com>
> Date: Fri, Mar 19, 2010 at 10:44 AM
> Subject: Bizarre results running Hudson job + CI
> To: CI4Py <testrun at codespeak.net>
>
>
> Hello !
>
> Before asking somewhere else I wanted to tell you about an issue I
> consider a bit strange . I've setup a buiild job @ testrun.org [1]_ to
> test Trac XmlRpcPlugin . It doesn't use a repository 'cause source
> code is checked out from a MQ repository hosted by Bb . That also
> means that I had to hack a little until I found a way to work around
> this.
>
> BTW, it would be very useful (for me ;o) in order to simplify build
> steps, if MQ extension could be enabled globally in build slaves . Is
> there any chance for this to happen, please ?
>
> Well, it finally worked , but the fact is that it is reporting
> inconsistent results AFAICS . For instance build 22 [2]_ displays 4
> failures + 19 errors . I ran the tests locally on a my PC and got
>
> {{{
> test_general_provider_error (tracrpc.tests.api.ProtocolProviderTestCase) ... ok
> test_invalid_content_type (tracrpc.tests.api.ProtocolProviderTestCase) ... ok
> test_valid_provider (tracrpc.tests.api.ProtocolProviderTestCase) ... ok
> test_content_encoding (tracrpc.tests.xml_rpc.RpcXmlTestCase) ... ok
> test_method_not_found (tracrpc.tests.xml_rpc.RpcXmlTestCase) ... ok
> test_resource_not_found (tracrpc.tests.xml_rpc.RpcXmlTestCase) ... ok
> test_to_and_from_datetime (tracrpc.tests.xml_rpc.RpcXmlTestCase) ... ok
> test_wrong_argspec (tracrpc.tests.xml_rpc.RpcXmlTestCase) ... ok
> test_xmlrpc_permission (tracrpc.tests.xml_rpc.RpcXmlTestCase) ... ok
> test_binary (tracrpc.tests.json_rpc.JsonTestCase) ... ok
> test_call (tracrpc.tests.json_rpc.JsonTestCase) ... ok
> test_call_permission (tracrpc.tests.json_rpc.JsonTestCase) ... FAIL
> test_datetime (tracrpc.tests.json_rpc.JsonTestCase) ... ok
> test_method_not_found (tracrpc.tests.json_rpc.JsonTestCase) ... FAIL
> test_multicall (tracrpc.tests.json_rpc.JsonTestCase) ... ERROR
> test_resource_not_found (tracrpc.tests.json_rpc.JsonTestCase) ... FAIL
> test_wrong_argspec (tracrpc.tests.json_rpc.JsonTestCase) ... FAIL
> test_xmlrpc_permission (tracrpc.tests.json_rpc.JsonTestCase) ... FAIL
> test_FineGrainedSecurity (tracrpc.tests.ticket.RpcTicketTestCase) ... ok
> test_getActions (tracrpc.tests.ticket.RpcTicketTestCase) ... ok
> test_getAvailableActions_DeleteTicket
> (tracrpc.tests.ticket.RpcTicketTestCase) ... ok
> test_getRecentChanges (tracrpc.tests.ticket.RpcTicketTestCase) ... ok
> test_attachments (tracrpc.tests.wiki.RpcWikiTestCase) ... ok
> test_getRecentChanges (tracrpc.tests.wiki.RpcWikiTestCase) ... ok
>
> [...]
>
> ----------------------------------------------------------------------
> Ran 24 tests in 45.674s
>
> FAILED (failures=5, errors=1)
> }}}
>
> My concern is because (AFAIK) the job relies on Py 2.6, nonetheless I
> get messages like this [3]_
>
> {{{
> 'Internal Server Error\n\nTracError: ImportError: No module named
> _multiprocessing'
> }}}
>
> :-$
>
> Could anybody please tell me ( what's happening | how to fix this ) ?
> I believed that multiprocessing was available in Py2.6 [4]_ and
> previous builds didn't say anything about this and the plugin doesn't
> use it (explicitly) at all ...


The version of python they have installed does not have
multiprocessing compiled into it. Without knowing the underlying
platform, and seeing the output from make when running the compile,
it's impossible to tell why.

jesse