USS Clueless              Voyages of a restless mind

Main: normal long no graphics Contact Log archives Best log entries Other articles Site Search

Stardate 20011222.1649 (Captain's log): I've got a new search engine installed for experimental purposes which seems to work pretty well. Unlike the one built into Greymatter, which is an on-demand brute force linear search of the entire site and thus slow, the new one crawls the site and creates a database. It takes about fifteen minutes to do that so I'll probably set that up as a nightly job (which would mean it would run at 4:00 AM my time, since that's when this version of Linux runs the daily cron tasks). The site currently is about 35M total, with everything, but a lot of that is the discussion system and a lot of it is graphics files. Still, there may well be fifteen megabytes of text here. The database it created is only 3M and it seems to be quite fast to access. Looks like a winner.

But as someone pointed out, one drawback of it is that it only shows five results on each page. More would be nicer. I was thinking about that, and so I reviewed the terms on this "free" code. See, it's GPL'ed:

This program is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation; either version 2 of the License, or (at your option) any later version.

And suddenly I realized that this is a very strong disincentive to making any change, however slight, to this program: because as soon as I do, I am legally obligated to become a redistribution site for it.

This is a considerable drawback, which had never occurred to me before I actually faced the prospect myself. One of the big arguments I keep hearing from advocates of open source is that "you have the source, you can make it do what you want and you can fix your own bugs." But you then also have to offer the modified version back to the world and pay for redistribution. So it's free -- unless you change it. Then it could become immensely expensive. Now I suspect that this particular provision of the GPL is probably more honored than obeyed, but if indeed everyone was conscientious about it then it would mean that the GPL charged developers but not users. This is not exactly a good way to motivate developers, wouldn't you think?

Of course, for the moment, there's an out: you can contribute your changes to some charity-run redistribution site such as SourceForge, rather than hosting it yourself and paying for the bandwidth yourself. But when VA Linux finally runs out of money (in about three weeks, at the rate they're going) and ceases to subsidize it, then that opportunity won't exist any longer. Will all those happy open source people out there who are merrily customizing follow through on their obligation to redistribute when it actually costs them out-of-pocket to do so? Heh. (discuss)

Andrew writes to me as follows:

Unless the GPL has recently undergone a huge change, you don't
have to distribute your changes UNLESS you distribute software. As near as I can tell, you have a web site. It runs software, but you don't transfer said software to anyone. Thus, the GPL doesn't obligate you to do anything.

Besides, as I understand the GPL's redistribution obligation, you're only obligated to give away the source of your changes to folks that you give the binaries to. Thus, the obvious solution is to not distribute binaries; just distribute source. (However, not distributing at all is the simplest solution.)