On 24 Mar 2010, at 09:28, Christopher Brown wrote:
> Can I ask a process point here?
Of course, any time.
> I'm seeing people saying "I may have missed a commit" and being asked to
> do a clean install to run these tests, which makes me wonder if we are
> all testing the same thing. When Noah sends out mail saying "these are
> the artifacts to vote on", is everyone ensuring that's exactly what they
> are testing? It doesn't appear to be the case.
I'm reading this as: Benoit thinks there's a specific cause for his test errors,
but he can't remember seeing a commit to code around that cause in the
past. Hence "missing a commit" as it not knowing it about.
Everybody in this thread must vote on the artefacts that Noah prepared
that is the same code for everybody (it comes with a gpg signature,
md5 and sha hash and voters should verify they all match).
Benoit, if my reading of this is incorrect, please let me know :)
> Before you write me off, hear me out. I'm an *avid* consumer of couchdb
> in my business today and vote at each release. I want to ensure
> quality, and understand the process.
When Noah sends out a mail with [VOTE] in the subject he links to a location
for a CouchDB tarball that is nominated to be the next CouchDB release.
Everybody on dev@ is encouraged to vote on the nominated release.
If you have any further questions, let me know :)
Cheers
Jan
--
>
> Thanks for your patience,
> Chris
>
> Jan Lehnardt wrote:
>> On 24 Mar 2010, at 02:44, Noah Slater wrote:
>>
>>> On 24 Mar 2010, at 09:40, Brian Candler wrote:
>>>
>>>> Anyway, nobody else seems to have had this problem, it's quite possibly
>>>> something to do with my setup, and browser-based tests are fragile anyway.
>>>> Hence I'm not going to vote against the release. It's a 0 from me.
>>> Not true, you are the second to report it in this vote.
>>>
>>> Any of the developers (Jan?) want to comment on how serious this bug is?
>>
>> Brian, Benoit, can you both double check and do a clean-slate installation
>> into a new --prefix if you haven't done so?
>>
>>> ** {{badmatch,{error,eacces}},
>>
>>
>> From Brian's stacktrace, I'm seeing the eaccess error which suggests that
>> CouchDB doesn't have write permissions to local.ini (or local_dev.ini, if
>> you run in make dev mode).
>>
>> --
>>
>>> [Wed, 24 Mar 2010 09:00:50 GMT] [info] [<0.113.0>] OS Process #Port<0.1864>
>>> Log :: function raised exception (new ReferenceError("map_funs is not
>>> defined", "")) with doc._id 8450bfd7a0371f80a9baf032553367f4
>>> …
>>> [Wed, 24 Mar 2010 09:00:50 GMT] [info] [<0.113.0>] OS Process #Port<0.1864>
>>> Log :: function raised exception (new ReferenceError("map_results is not
>>> defined", "")) with doc._id 8450bfd7a0371f80a9baf032553367f4
>>
>> These come from the view server and look like there's a weird condition.
>> Chris, maybe you can comment on this?
>>
>> Cheers
>> Jan
>> --
