Hello, Le jeudi 3 octobre 2013 13:13:10, Sandro Santilli a écrit : > On Thu, Oct 03, 2013 at 12:00:05PM +0100, Jonathan Moules wrote: > > Assuming it's not already done, I'd suggest that requiring Unit Tests for > > all new features should be mandatory for a Pull Request/patch to be > > accepted into QGIS. In the long term that would hopefully help alleviate > > the significant number of regressions that we're seeing in 2.0. > > +1 > > Of course providing tests will only be effective when there's also a rule > of not accepting changes beaking any of the tests, and that's only > effective when there's a continuous monitoring of testsuite runs. All of > that is likely yet to be defined by the new Testing and QA Manager.
And another +1. It's time to focus on code quality to reduce bugs and maintainance costs, for which it is always difficult to find funding. We need more unit test, more code quality dashboards and much stricter rules relative to what code has to be accepted into master. A PostgreSQL-like code inclusion workflow, with commitfest and review, could be something interested, to be discussed. On the funding side, for information, public administrations, at least in france, cannot pay for a time based contract. Estimating how much time a bug fix could take is a really hard task (except if you fixed the bug already). This is a problem both for company willing to fix bug for paid contracts and for public administrations wanting to finance bugs, and can explain the funding situation related to maintainance and bugfixing. Vincent > Speaking of which, the psc page describes the roles but there's no link > to the currently appointed person for each role: > http://www.qgis.org/en/site/getinvolved/governance/organisation/psc.html > (nor links to contextualize the path that brings there from the homepage) > > --strk; > > > Jonathan > > > > On 3 October 2013 11:51, Régis Haubourg > > <regis.haubo...@eau-adour-garonne.fr > > > > > wrote: > > > > > > Hi Matthias. agreed. > > > How to fund, this is the question. I have budget (at least at the > > > moment). I haven't been able to spend it completly in 2.0 release > > > sprint because all goes too fast for classical contract, where I'm > > > constrained to pay for a specific work (debug or feature). > > > Please be warned to public finances are not good in europe, and every > > > unconsumed budget can be erased. > > > Sponsoring should be a way to finance infrastructure consolidation, but > > > in current rules, sponsoring can't be oriented. > > > > > > I need help from the community to have a serious real use case unit > > > test suite. I'm ready to fund a part if PSC gives an infrastructure > > > and a manager. If no work canvas is given, I'm sure I won't have any > > > successfull commercial proposal to such a funding. > > > > > > I also kindly ask other funders to systematically include unit test > > > developpement for every new feature. > > > > > > Again today, I found 3 regressions, hard to find. I thought once 2.0 > > > was out, I could spend less time testing, and more time working. It's > > > still not the case. > > > > > > Again, you gave a lot unpaid work, when I was struggling to find a way > > > to spend some.. > > > Hope we'll find a way. > > > Régis > > _______________________________________________ > Qgis-developer mailing list > Qgis-developer@lists.osgeo.org > http://lists.osgeo.org/mailman/listinfo/qgis-developer _______________________________________________ Qgis-developer mailing list Qgis-developer@lists.osgeo.org http://lists.osgeo.org/mailman/listinfo/qgis-developer
