One point which may be relatively easy and valuable is to get Django to support backend-provided tests, and do the final clean-up of backend-specific tests from the general test suite. That is, put all the postgres-specific tests somewhere in django.db.backends.postgresql_psycopg2 etc, and have the Django test runner pick them up from there, only if that backend is actually specified in settings. There are some open questions about this -- e.g. proper handling of multiple databases -- but in general, this will both make it easier for all backends to pass the Django test suite, and make it easier for each backend to improve its own quality.
On Tuesday 10 March 2015 04:46:28 Yichun Duan wrote: > Thanks a lot for your suggestions. I'll try. > > 在 2015年3月10日星期二 UTC+8上午8:10:49，Russell Keith-Magee写道： > > > Hi Yichun, > > > > On Mon, Mar 9, 2015 at 4:13 PM, Yichun Duan <[email protected] > > > > <javascript:>> wrote: > >> Hi, > >> > >> I'm Yichun and I'm interested in the Project ' Improving the less > >> popular database backends' in GSOC 2015. I major in computer science in > >> Peking University. I've worked with C++, Java, Python and I've written > >> several webapps using Django. > >> > >> I have some questions related to this project and your answers will be > >> of big help. > >> > >> (1) I'd like to make sure that I've understood this project in a right > >> way. For example, if I choose to improve the backend of Oracle, then I > >> can choose some related issues in Trac's list of Oracle issues > >> <https://code.djangoproject.com/query?status=assigned&status=new&summary > >> =~oracle&or&status=assigned&status=new&keywords=~oracle&col=id&col=summa > >> ry&col=status&col=owner&col=type&col=component&order=priority> and > >> prepare to work on them during the whole term, right? I've not seen a > >> mentor for this project in the GSOC guide, and I think someone here may > >> help me. > > > > I don't think anyone has specifically stepped forward to mentor an > > Oracle-based project, but if you put forward a strong proposal, we will > > find someone to mentor you. > > > > As for what to tackle for the project itself - one approach for your > > project would be to just tackle a collection of Oracle-related tickets. A > > better approach (more likely to be accepted) would be for your to analyse > > the tickets that have been reported, and see if you can find a common > > theme (or themes) - and then propose a way that we can eliminate those > > problems at a higher level. > > > >> (2) Actually I'm familiar with MySQL and SqlServer, but never use Oracle > >> before. Will it be a big obstacle in my work if I choose to write the > >> backend of Oracle? Or if I begin studying Oracle right now, it won't be > >> a problem? > > > > A history of Oracle experience would definitely work in your favour, but > > *not* having that experience won't necessarily work against you. The fact > > that you've got experience with multiple SQL flavours means you should be > > aware of some of the interesting ways SQL is interpreted. Highlight that > > experience in your application, and that should be enough. > > > > Alternatively, you could try to work with the developers of the SQLServer > > backend and see if you could improve support for that backend. > > > >> (3) I'm a new contributor in Django. Before I start working on my > >> project, should I start from fix simple bugs or something else? > > > > You don't have to, but it would certainly be looked upon favourably. In > > particular, if you were to tackle one or two small Oracle bugs, that > > would be very helpful to prove that you're going to be able to quickly > > pick up the new skills you require for the GSoC term. > > > > Yours, > > Russ Magee %-)
