On 15/06/11 at 10:32 +0200, Christoph Berg wrote: > Re: Lucas Nussbaum 2011-06-14 <20110614172545.ga8...@xanadu.blop.info> > > UDD was importing DEHS data from > > http://dehs.alioth.debian.org/udd_dump.txt > > abi-compliance-checker|1.21.12||1.6-1||1.6-1||f||f||1970-01-01 00:00:00| > abicheck|1.2||1.2-5||1.2-5||t||f||2011-05-04 06:15:58| > abind|1.3.0||1.3.0-1||1.3.0-1||t||f||2011-05-04 05:47:03| > > wtf. > > > Could you export your own data using a similar (preferably identical ;) > > format over HTTP? > > Which fields do you actually need from that?
In my notes: Query used (on DEHS' side) to generate the data: SELECT COALESCE(u.name, e.name) AS source, u.up_version AS unstable_upstream, e.up_version AS experimental_upstream, u.version AS unstable_version, e.version AS experimental_version, u.dversionmangled AS unstable_parsed_version, e.dversionmangled AS experimental_parsed_version, u.updated AS unstable_uptodate, e.updated AS experimental_uptodate, u.watch_warn <> '' AS unstable_failed, e.watch_warn <> '' AS experimental_failed FROM (select * from pkgs where dist='unstable' AND watch IS NOT NULL) AS u FULL JOIN (select * from pkgs where dist='experimental' AND watch IS NOT NULL) AS e ON u.name=e.name WHERE (u.dist='unstable' OR u.dist IS NULL) AND (e.dist='experimental' OR e.dist IS NULL) Ideally, I would get the same data. > The raw data looks like > http://qa.debian.org/cgi-bin/mole/watch/.dump/watch > > Pick some fields, writing the exporter isn't too hard then. Also, CSV > without column headers is ugly, what about something yaml-ish? > > package: > upstream_version: xxx > date_checked: xxx yes, YAML would be perfect. - Lucas -- To UNSUBSCRIBE, email to debian-qa-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/20110615084236.ga13...@xanadu.blop.info
