Aidan Delaney’s weblog

So it seems that I just _can’t_ keep a diary, even if it’s in electronic form . I’ve been busy lately
* helping revive NIS (took a few days),
* writing host program for the fl86 development board
* reding about UNIX/MULTICS/Plan9 and a host of other OS’es

I’m taking a cruise down the canal to the Shannon this weekend with Lori, her sister and her Dad. Should be fun. And next week we have another informal get together of the free software enthusiasts. I’m off home now to put the finishing touches on my host program. I’ll post source tomorrow.

I’ve come up with a working definition of an operating system, that “An OS should provide everything needed to run a shell”. Its a bit all encompasing, but at least I can refine it now.

I had an interesting conversation with Bart at lunchtime about the semantic differences between using the traditional UNIX/Plan9 file interface for accessing global namespaces, and treating these namespaces as objects of type file. I’ve yet to come up with a conclusion, but it’s interesting.

Charles is building me a computer from a ZX80A, his favourite microprocessor. It’ll have a keypad for input, an LCD for output and a huge 32 kilobytes of RAM. Should be ready in two weeks. I need it to play around with the practical implementation of OS’es.

I met with Tom today and showed him my game plan for the next few months. It basically means that I’ll be doing a lot of OS research alongside the theoretical research. I also get time to poke through source code. I’m going off to search the InterWeb thingie for useful resources.

I’m required by my contract with the university to do a few hours of work per week. As part of that work I take care of the Tass website. I just relaunched it, and it can be found here http://tass.cs.may.ie. It’s not fully finshed yet, but it looks good.

The fsfe-ie WikiWikiWeb is now operational at http://beecher.net/scripts/phpwiki/index.php/HomePage.

I’m reading some research on multilevel generative descriptions of languages today. Exciting stuff, but we can’t all be kernel hackers Some of us have to do basic research and to be honest I’m quite enjoying it (and have been for the past 8 months).

My mathematical model is similar to the code that implements it (in fact the code is just an alternatate denotation of the model). I’ve been playing around with the model for the past few days to no avail. Today I took a big picture view of it and simply looked at the high-level mathematical properties of the system. This allowed me to come up with a better solution to the problem that originally showed the bug in my model.

I had a solution to the mildly context sensitive language ww. This solution didn’t work on the model as there was a bug in said model. I think I fixed the model, but I tried the same solution to ww. Given that the model had changed, it was probably not a good idea to use the same solution. So I’ve come up with a better high level solution. I’m going to see if my “fixed” model (which is still slightly broken) can solve it. Then fix the broken bits in the most simple way and finally check the model for consistency. I don’t want to get bogged down in the actual implementation of the model again.

On another note the first free software foundation Europe – Ireland (fsfe-ie) meeting last night was interesting. Strangely enough 4 of the people there worked in TCS. We’re hopefully going to get a good anti-patent campaign off the ground soon. Imagine if I was able to patent the code that implements my model even though I know the code is just an alternative way of writing the maths, we’d be on our way to allowing the patenting of maths then. Not a good idea

I’m heading to the fsf-ie meeting tonight, anyone is invited, 7:30 in messrs maguire. Mags also updated her campaign website with a press-pack, it’d be nice to see her get her .ie soon, but her current site’s at http://www.redbrick.dcu.ie/~afrodite/E-Voting/. I’ll have to find out later why she switched to using RedBrick instead of MiNDS>.

Fixing the “bug” in the mathematical model has turned out to be more perplexing than I originally thought. I’ll have to talk to my supervisor before the end of the week, in-order to get it all straight in my head again. The problem is with the “malloc” functionality. It sounds as if I’m describing a LBA now and not what I originally intended.

I’m working on a version of malloc for TPA’s. It’s needed now because of the model change.

I’ve been busy lately. Went to the Eminem gig on Friday, it was a load of crap. I’ve seen better garage bands. I tidied my desk (big job) and yesterday I think I fixed most of the problems with my mathematical model. Today I’m in bloody early trying to getting a Solaris box to behave. I have no love for Solaris.

Later I’m going to have a look at the differences between traditional CFG concatenation operators and the operator defined in my model.