I blogged about the new Harvest a couple of times already. Yesterday we released 0.2.4 which fixed a bug which was filed more than a cycle ago… it merely said: “import json”.
The way Harvest works is that it reads a list of URLs that list opportunities, so Harvest is very easily extensible. It doesn’t need to know about bug reports, merge requests, upstreams, downstreams and the like: it just knows about packages and opportunities.
The old format for opportunities was:
<sourcepackage>,<url>,<description>
It was only a matter of time that this wasn’t enough any more. Some people suggested to add something like “long description” or “severity” to opportunities to make it clearer what they are about or to make an informed decision. So we moved to JSON where it’s very easy to add optional arguments and you don’t have to guess arguments based on their position.
If you wrote a script that spits out .csv files, that’s fine - it’s still supported. In the new world you’d do it like this:
[{"source_package": "xine-lib", "short_description": 123456, "link": "https://launchpad.net/bugs/123456"}, {"source_package": "zope-archetypes", "short_description": 54321, "link": "https://launchpad.net/bugs/54321", "description": "Please sync zope-archetypes 1.3.9-2 from Debian unstable", "severity": "undecided"}]
Only “source_package”, “short_description” and “link” are required. Everything else is optional.
I’d LOVE to see more opportunities for Harvest coming up. Once you got your script set up, just bzr branch lp:harvest-data add the URL to the JSON output, and propose a merge.