On Wed, Apr 11, 2012 at 7:55 AM, "Paweł Hajdan, Jr." <phajdan...@gentoo.org> wrote: > On 4/10/12 11:17 PM, William Hubbs wrote: >> On Tue, Apr 10, 2012 at 10:45:14PM +0200, "Paweł Hajdan, Jr." wrote: >>> On 4/10/12 7:54 PM, William Hubbs wrote: >>>> I can name a couple of issues that are api limitations that we can't do >>>> anything about: >>>> - you can't search on cc: or keywords fields. >>> That's going to be a problem for arch testing needs (e.g. STABLEREQ >>> keyword and x86@ cc-ed). How about doing the searches "the old way" that >>> allowed the above. >> >> That is not so easy to do since we have completely gotten rid of the old >> method of communicating with bugzilla. That method was not reliable and >> had broken several times with bugzilla upgrades, but using the web >> services will be more stable. > > Right, scraping HTML that way was obviously brittle, but at least it > worked (in the pragmatic sense). Without that functionality, the > batch-stabilization tools I and other devs use would be broken. > > If possible, please restore the old code, possibly marked as deprecated > and not officially supported (best effort, patches welcome is fine), and > named maybe search_brittle or something similar. >
Do you need the command line front end for your arch testing utilities, or are you using the bugzilla.py module directly? If the latter, I think it would be possible to restore the old bugzilla.py module, possibly with a different name. If you actually need the front end (bugz and cli.py), it would probably be easier for you to just have two copies of pybugz: an old version for searching and a new version for making changes.
