Wow, that’s a lame title. I couldn’t think of anything better 
This is a general-purpose update as to what I’ve been doing of late, so no single title seemed appropriate, except “What Dave’s Been Doing Lately” (and I think the title I finally went with is more punchy).
Read on for general babble about what I’ve been doing lately, and a random thought about the best way of getting feedback about server development.
Recently I’ve been doing a fair bit of reorganising behind the scenes. And I don’t just mean mb_server coding – if it were that, you’d see evidence of it in the mb-commits list or over at the CVS viewer page.
No, the stuff I’ve been up to is more along the lines of shuffling things about on the server itself. Upgrading Apache. Improving the efficiency, placement and content of Apache’s log files. Fixing little MRTG-related problems. Putting it in (a private part of) CVS. Anyway all these things help to keep things running smoothly, and I don’t often talk about them, so I can see how this could mistaken for inactivity. And correcting that mistaken impression is the whole reason this blog was created.
Also the other day I pruned the TRM database again, for all the same reasons as the first time – the TRM database grows, outstrips the available memory of the machine, and then slows down dramatically, taking the main MusicBrainz database with it. Read the above article for more details, and for pretty graphs illustrating it all.
Finally, after a few weeks of hard disk woes, my machine at home is up, stable and generally happy with its life, which means that I can get on to usefully hacking on mb_server again. I think the annotations work is sufficiently complete to be merged into the trunk soon. I haven’t implemented all the requested features – I think it’s more important to get this out into the wild first, 95% complete as it is, and work on the last 5% some other time.
Which brings me on to my final point …
I don’t actually get an awful lot of spare time to spend on this fine project. Currently I’m probably averaging about 30 minutes per day during the week, and a bit more most weekends. As a result, I’m quite conscious of the various less-than-productive MB tasks I spend that time on (instead of useful stuff like server administration or mb_server hacking). Editing and voting have gone right out of the window, I’m afraid; but that’s OK, because effectively no harm comes from me not doing that any more. Reading the mb-users list is the biggest drain by far, but to a certain extent I can control that by simply not reading all of the posts – which is why, as I said in a post I made the other day, I’m far from up-to-date on all the latest “Style” thinking.
However, the one thing that does take a lot of my time, and which I haven’t yet found a way of dealing with, is collecting and collating feedback on server development. For example, when I threw the “Annotations” work out for testing recently, lots of feedback was forthcoming – which was great! The trouble is that that feedback comes in the form the rambling tree structure which mailing lists tend to follow, which means that I often have to go digging around the list to try to collate everyone’s ideas, ignoring the irrelevant postings (that is, the ones which are irrelevant to the final outcome), and work out what the consensus of opinion is. Only then can I go and take that consensus and act on it.
I’m wondering if a better approach would be something more Wiki-based. Or maybe someone would like to be a nominated spokesman of some kind, responsible for tracking the thread, trying to get the various participants to reach some sort of agreement (if there is any disagreement, that is); then making the results of the thread available to me in a nice short, concise, accurate, non-rambling form.
In other words, the complete opposite of this blog posting 😉
Anyway, your thoughts are, as always, welcome. I’m off to go and hack on the server now, until the next Real Life(tm) interruption.
MetaBrainz Blog 