Re: New upstream data in blends query
Hi Andreas,
Andreas Tille <andreas@an3as.eu> writes:
> On Sun, Mar 06, 2016 at 10:28:34AM +0100, Ole Streicher wrote:
>> > We've seen some more and brighter ones but here the foreground
>> > is nice as well. ;-)
>>
>> Whow! This is really worth going there!
>
> Yes. And even if I'm back home, I'll remain in Iceland mentally for
> some while. (And for Debian issues I need to work down a certain
> backlog so please be patient with me.)
No problem, I will come with my problems slowly ;-)
> Just one thing: I've get a cron job message:
> Traceback (most recent call last):
> File "./generate-debtags-field-list", line 29, in <module>
> user='public-udd-mirror', password='public-udd-mirror')
> File "/usr/lib/python2.7/dist-packages/psycopg2/__init__.py", line 164, in connect
> conn = _connect(dsn, connection_factory=connection_factory, async=async)
> psycopg2.OperationalError: could not connect to server: Connection refused
> Is the server running on host "public-udd-mirror.xvm.mit.edu" (18.181.6.246) and accepting
> TCP/IP connections on port 5452?
This may be a bit older (on saturday there was some issue in the
connection), and it is for sure not connected with my changes: I did not
touch the connection string at all (except locally for my tests), and I
just re-checked that I didn't accidentally pushed that to alioth.
So, this is different. Since generate-debtags-field-list is written and
maintained by Petter, maybe he could say something.
>> OK, I now checked in quite a number of changes (in the python-2
>> directory). blendsunicode.py and to_unicode() are gone completely
>> there. This is checked locally, and after the afternoon rebuild I will
>> see whether there are still problems in any blend.
>
> Nice!
After a few iterations I found the last issue with some edu tasks
today. Now, the whole thingy works without to_unicode() and without
encode errors. And with correct encoding (as far as I could see it)!
>> For the python-3 directory: would you mind if I would move the work from
>> a directory into a separate branch? This would make taking over changes
>> from the python-2 directory much easier (cherrypicking).
>
> I'll keep Akshita in CC. I'm for the solution that causes less work -
> so I'm fine with your sugggestion.
I'd revoke it. I tested this yesterday locally, and I couldn't manage to
let changes easily flow from webtools to the Python-3 version: the two
branches do diverge too much already.
So, let's keep them as they are, and see where we go. In the moment, I
think it would also be not too difficult to slowly make the original
subdir compatible with Python-3.
....
Now, may I come with one first problem? :-)
The upstream_repository field does not give a value at all, and the
upstream_bugs field works only in a few cases: it works only if the
"Upstream-Bugs" field is in the first line of d/u/metadata, so the
parser seems not to read the whole file for them. I think this is a
problem of the UDD side.
Best regards
Ole
Reply to: