Older blog entries for AlanShutko (starting at number 11)

Learning Metafont and CWeb lately. Along with that, I've been beating more on mmm-mode to support CWeb. There are a number of areas where mmm-mode doesn't do things the best way for cweb, so I'm doing a bit of hacking on its internals. CWeb looks like a good choice, because it's different enough from what mmm-mode handles already that by the time I get a submode class for it, mmm-mode should be able to handle almost anything easily.

Shipped off some quick patches for mmm-mode. Now it works enough to see that it works, but since I don't actually use the languages it has built-in support for (besides RPM specs, but that's broken slightly by Emacs 21) I haven't used it much. My copy of the CWeb book should be arriving today (along with the Computers and Typesetting boxed set!) so I'll be putting together some support for that soon.

mmm-mode looks very promising. It's more sophisticated than multi-mode.el, which is good and bad. It doesn't have to switch between modes all the time like multi-mode, but it can't yet handle cases where you add mode delimiters without using its interface (it uses overlays to keep track).

Other news, updating my resume. Anyone looking for a guy on Long Island?

Ok, useless weekend. Needed to clean things way, way too much. So finally got rid of the big plasticky green thing with the glowing lights, moved the fish so the cat could torment it in a new place, moved around furniture and vacuumed lots. Nothing fun got done this weekend.

jmg:

I agree that tag translation should be done as part of the end app, but in my case my marc convertor was just an offshoot written using code from what may someday be the end app, so it was trivial to make it translate the tags. It would be nice if we had a common file format for the tag info... I have a simple format that I load in to do the mappings, but now that xml libraries are more widespread, it should probably be converted. There are just so many fields, subfields, indicators that it's pointless for every project to have to translate them, and most end up never translating them at all.

Ah, it's the weekend. Time to relax and figure out why my wireless access point periodically goes crazy and requires a reboot. Well that's the problem of playing with new toys, but it's really nice to wander around cleaning with a laptop near me at all times.

jmg:

I'm not sure why you think using spaces instead of tabs will make source more difficult for someone else to read. At worst, it'll make them read it exactly the way you wrote it. How is that bad? If they don't like your style, they'll have problems anyway... there are more differences between C styles than tab stops.

If you only use TABS at the beginning of lines, you'll probably do alright with people using different tab widths. But sooner or later people will want to line up things which come after text... maybe variable names in declarations, maybe comments after code. That's when tabs are a big problem, because what lines up for you at 8 spaces per tab may suddenly not line up at 4 spaces per tab. Spaces don't have that problem.

Funny thing, I also have a MARC binary->ascii converter. I wrote mine as part of a larger project to make an EndNote-like app for X. My MARC converter also can display human-readable names for fields and subfields, btw.

My project has been put on the back burner, but if you're interested in something like that maybe we can get it off the ground.

Wikis don't seem to do all the stuff I want. First, AFAIK they require you to check the wiki all the time to see new stuff. This isn't bad when you're building a knowledge store, because then you can find new information when you need to look at it, but it isn't as conducive to a conversation.

Mailing lists are good because people can forget about them when they're quiet, but when someone posts, you're immediately reminded. You have big flags showing when someone is asking a question... you don't have that built-in on a wiki.

So, I think there are definate places for mailing lists. They haven't been replaced by wikis, and I don't think they will be. They serve different needs.

But I'm not just interested in mailing lists. Mailing lists are everywhere, and they're pretty easy to put up. What I'm interested in is the egroups infrastructure. It's insanely easy to put together a new mailing list, for one thing. But once you do, you get all sorts of tools:

• A file area.
• A calendar, which would make it even easier to schedule things than a mailing list.
• Little flatfile databases you can use, with predefined schema for address books, recipes, FAQs.
• Polls, included automated shutoffs, result counts, etc.
• Automated mailing of faqs and such
• A chat room

You get all this stuff for every list, without having to put the pieces together yourself. Now, most lists I've seen don't use all that, but it's there, and easy to use if the group decides to. And you can take advantage of everything without having the knowledge to set it up, and they all have pretty decent interfaces.

It's just like sourceforge. Sourceforge gives every project tons of tools, which they may or may not use, but they're more likely to because they don't have to develop or install those tools themselves. And the release of the sourceforge code means you can set up your own targetted project groupings on your own machine.

I want that kind of thing for free software mailing lists. A wiki isn't a replacement for all that, although it would certainly be a good tool to add to the list. (After all, people could want their own wikis for their groups... less wikiname conflicts.)

Checked in some minor fixes to gweather I've had hanging around for months, waiting for CVS write access.

Also playing with starting a list on yahoo groups (nee egroups, nee onelist). They really have a fairly nifty setup with lots of features. Things like the built-in flatfile db, the autoposting, file area, calendar and polls are pretty cool, and I could see them being very useful to a community. Unfortunately, egroups has two major problems:

• Ads, which can be fixed by paying some money
• Lack of control, which can't.

It would be neat if there were an open-source system which were similar. Right now, if you wanted to set up an in-office set of lists, you couldn't... you wouldn't want all your private info going through egroups and being archived outside your company. Or maybe you're a large institition like a university which wants to set up a tailored set of lists, give some groups more file space, allow others to run CGIs, etc.

I'm thinking something like the sourceforge software, but with the features from egroups. (Maybe even the sourceforge software: I don't know how many mailing lists need a CVS repository, but many projects would benefit from polls and the calendar.)

Of course, with my CFT I'll probably not get around to it for a few years... maybe someone else is interested?

Ok, so it never did inspire me. Things have just been a bit busy. I've done some stuff for home (updated my envelope printing cgi, came up with a logo for our apartment, more maintenance on album, which needs a new name since freshmeat has a project named that way) but nothing that should see the light of day. It's terribly hacked up, and needs massive reworking.

Yes, I know open source will fix it, but it's a matter of pride. I could fix it, and I don't want a potential employer to look at it and think that's all I'm capable of. (It's just all I'm capable of when letting a prototype grow into a monstrosity.)

But I'm looking into literate programming and I'm going to redesign my toys (whether they end up in cweb or something more universal is yet to be determined).

Gosh, it's been a while. I'm no longer a Corel Slut, but I still haven't had much time for hacking. The weather has been, overall, pretty good and I've been spending more time outside doing fun things than working on code. And other hobbies have begged for time. But now I have a new Thinkpad A20p and getting everything working under Linux has inspired me, so I may actually code something! Try to fix up some of the projects I've had lying around.

Anyone interested in working on a bibliography/reference manager?

Not much hacking lately... Corel's been making a bunch of releases, and as an official Corel Slut new stuff sucks up some time.

Ok, this is a test entry to see what's up with Kai's W3 problems... hmmm... I dunno. Let's see & <if> w3 mungles this.