(I think) I am a whopping one bug away from releasing MetaFS 0.1.4. This version should (finally) be able to serve homedirs to KDE, and will generally be safer and more bug-free, but it will be slower and sort-of-leak memory. (Blame FUSE, not me. :p Well, OK, blame me but realize these issues should be fixed in 0.1.5 when I stop using their annoying userspace library.)

Most of my work with MetaFS lately could be categorized as "putting out fires" ... or, less cynically, cleaning/tightening up the API, cleaning up the library internals, trying to handle more and more error cases, etc.

But I'm still disturbed at the number of stupid mistakes I've already found. Logic errors, semicolons at the end of if-tests, ... the list goes on. This says to me I need to (a) stop coding so much at night when I'm more tired, and (b) write a lot more regression tests to more thoroughly exercise the API. The tests I have so far check for basic, basic functionality ... but they don't really stress the API. They don't try to do anything illegal, and there are some aspects they just don't cover (like readdir(), until very recently).

I also need to go through and make sure the documentation hasn't gotten out of date. I've tried to keep up with it, but I caught some old cruft in there today that didn't apply anymore, and pulled it out.

::sigh:: If I keep finding shit like this, I'm going to start getting worried about accomplishing my goals for Summer of Code. Speaking of which, I still haven't heard one way or another, and based on the rumours I've been hearing on the mailing list and IRC, I probably won't know until Friday. Dammit, I want to know NOW. ;P

-- Des