Date: Sun, 1 Sep 2013 22:56:56 +0000 From: Matthew Windsor <mbw500@york.ac.uk> To: soc-status@freebsd.org Cc: Justin Edward Muniz <jmuniz@freebsd.org>, Eitan Adler <eadler@freebsd.org> Subject: GSoC Status: Week 11 Message-ID: <CAFxS2Cir5QMU2xKYbGpxSu5jLyEUs%2BBnEzv2stoNw%2BO=95Q_pA@mail.gmail.com>
next in thread | raw e-mail | index | archive | help
Hi there, I haven't got around to making the functional tests I was planning for this week. Most of the work I *have* managed to get out has been in the form of more unit tests, with a hope of covering as much of the code that doesn't touch the backend as possible. (It seems that I can't get any code that does to compile, as it wants to link against the backend, and the backend is what the shared object slots into.) The Makefile directives for the tests are really sloppy, so I'd like some direction on how to make the tests in a way compatible with the BSD makefile system (I'll probably roll off another email to this effect), but for the moment they're building adequately-ish and proving to be a useful aid in testing. Most of the unit tests cover small utility functions at the moment; I'm hoping to extend this. Something the unit test angle has caused me to do a lot of in the code is deliberately structure things more and more so the backend-touching aspects are factored out as much as possible, and as much code as possible is left without access to the backend. This is a big change from the way I was originally writing the code (touch backend as soon as possible) and will likely influence the way I reimplement error reporting. This does unfortunately mean I'm cutting it fine with calling the backend code-complete and ready for proper QA... One known outstanding issue that I want to fix ASAP next week is that listing repos will show idents and names in the wrong place. Indeed, the query system now uses repo names (repo-packagesite) instead of idents (packagesite) as this is much simpler to implement (I can just pull the repo name from the package). This could be done as a functional test, too. As a minor note, I've pushed the C standard back down to C99. This should allow FreeBSD stock gcc to compile the backend. In summary: I did unit tests instead of functionals this week; code is still in flux which might make testing harder; C99, not C11. ~Matt
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?CAFxS2Cir5QMU2xKYbGpxSu5jLyEUs%2BBnEzv2stoNw%2BO=95Q_pA>