[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Developer.php: bugs status



* Jeroen van Wolffelaar <jeroen@wolffelaar.nl> [2004-03-06 16:39:48 +0100]:

[I don't know if you subscribed to the list so, don't blame me if you
receive this mail twice ;)]

> Hi,

Hi Jeroen,


> Two requests regarding proccess-index.pl:
> - Igor, sometimes a lot of packages have no bugcount. This is because
>   you directly do process-index.pl > bugs.txt, in stead of
>   process-index.pl > bugs.txt.new ; mv bugs.txt.new bugs.txt

You are right but these days there are some disks space problems on
klecker too. I have fixed the crontab entry.

>   If you do the to-dba processing also in process-index.pl (by external
>   calls) and make them http-accessable, also other scripts and such can
>   use your output. Currenlty, a lot of such stuff is done multiple
>   times, as you admit in the code for example, the PTS code does about
>   the same too.

The to-dba process is done in cvs-qa/data/cronjobs/ddpo.bugs
---> cut <---
# Need a version of libdbX-util compatible with the libdb version used by php4
# Fortunately the versions on klecker are compatible
rm -f bugs.db.new ; awk -F: '{print $1; print $2}' < bugs.txt | sed 's/\\/\\\\/g' | db_load -T -t btree bugs.db.new; mv bugs.db.new bugs.db
---> cut <---

> - bugs.debian.org updates bugs in intervals. The current 'bug info last
>   updated' is meaningless, because it cannot detect any failed script in
>   the process alongway. To fix this:
>   * at the beginning of process-index.pl, extract from somewhere the
>     real last run of the update-BTS process. If you're running from a
> 	BTS mirror (as I believe is the case), take care this is correct :)
>   * Add to bugs.txt simply 'BTS_TIMESTAMP: unix_time(last-BTS-run)'.
>   * Fix developer.wml to read BTS_TIMESTAMP's timestamp, and not
>     bugs.date (I could write a patch if you want)

I extract the information from /org/bugs.debian.org/spool/index.db on
master that is not a mirror. Moreover the index.db points to
index.db.realtime
I can extract the index.db timestamp that is significant enough for our
purpose.

Thanks a lot Jeroen ;)
PS: I didn't miss you proposal ;)
-- 
Igor Genibel 
http://www.answare.fr/                             igor.genibel@eds.com
http://www.tuxfamily.org/                            igor@tuxfamily.net
http://people.debian.org/~igenibel/                 igenibel@debian.org
GPG: 1024D/1648F6C8

Attachment: signature.asc
Description: Digital signature


Reply to: