Matthew Walton’s Blog

Saturday was an unusual day. Instead of bumming about as I usually would, maybe getting some programming done, or some gaming or writing, I got up obscenely early and went with my parents to Mum’s work — Bourn Hall Clinic — where they were having a party to celebrate the 25th birthday of Louise Brown, the first baby conceived using IVF to be born in the world. You may have seen something about this on the news.

Miss Brown was there of course, but not until later. We got there at nine o’clock in the morning, and there was much to be done…
(more…)

Well, today came a new heatsink and fan and case for my computer. In the quest to stop it over heating, these seemed sensible items to get, although admittedly the case is rather an indulgence. Read on for my impressions (and a few pictures) of my new Lian Li PC60 case and Thermalright SLK-900U heatsink.
(more…)

Two interviews done now, both thanks to ECM — thanks to Jim. Also thanks to the companies of course!

Now the nail-biting wait to see if I get called back to either of them. Feedback from the first says I was very bad at the start of the interview but very good at the end, when I would have suggested the other way around. Shows my lack of experience with this sort of thing I guess. Hopefully I didn’t do anything like that today!

I’m now stuck in a curious kind of reluctance to hear how I did, because in this state I can pretend I did well, but when the news comes it might be bad. There’s no choice but to reach for the chocolate, I suppose.

Is ‘complicatedly’ even a word? No matter, it is now.

Anyway, I’m going along writing Towel’s new playlist code, and I realise some of the loops I’m going to have to jump through to keep the data within the playlist structure (which is like a tree kind of thing) synchronised suitably. I’m storing some information at each node rather than recalculating it each time it needs to be accessed, because it will work out alot more efficient that way. I don’t like to think of what kind of mess it would make if I had to recalculate some of this stuff every time the song changed. This data is needed to make random track selection reasonably fair, because the nested-group concept means I need to weight the random algorithm at each level so it would be more likely to go down a branch with 16 tracks in it than stop at the current level which contains only 3 tracks. Of course, it might do that anyway, but that’s what a fair random picker is all about.

I’m a little worried that the random algorithm might work out being a bit slow, actually, but short of implementing my own TreeModel, which I’m not about to do, I can’t really do much about it and keep the playlist working the way I want to.

Anyway, for now it’s moving along. Need to add some more data items for things like the current track, so it can keep tabs on it, and the code for next-track selection of course. Then we can finally plug it into Towel’s main code and make it play music. Jonathan’s nice GStreamer backend will be doing that of course, it already works with the old playlist. In fact, the playlist code doesn’t touch it at all, save for a little bit of use of his Towel::File and Towel::Loader classes to do filetype detection when a track is added to the playlist.

Also need to work out an efficient way of loading the playlist. Ideally I need to build the tree then calculate the metadata just once, rather than with each add, so loading from playlist files will need a special code pathway… thankfully a generalisation of the current one should accomplish that.

All fun though

Things are moving along quite nicely right about now. Got a few interviews lined up through my recruitment agency, wonderful people that they are. Even if I’m not successful, I’ll tell myself it’s valuable experience. And it’s more than I’ve had before, so at least someone’s interested. Eventually there won’t be any better candidates and someone will have to employ me. Wheeeeeeeee!