Uni

This afternoon I have successfully defended my thesis on "Game semantics for object oriented languages".

I'm rather happy about that.

Gulp

Happy new year.

We didn't have a white Christmas, but I am having a white thesis submission day! Rather appropriate, as I have been looking forward to this moment for a long time. Of course I'm not really finished, in that my thesis is as yet unread, I have to have my viva voce, make corrections and then hopefully be awarded a PhD. But I've handed in a huge stack of paper to the appropriate people, and I'm finished for now.

I had a nice relaxing few days up at my mothers, seeing family and enjoying a lovely Christmas dinner (as always). Since then I've been doing not much at all, mostly playing games. I bought a new PC for gaming as I finished my thesis, and I've mostly been playing Portal (sheer brilliance), Team Fortress 2 (addictive) and a bit of Hellgate: London (shiny). I've still to get into Oblivion and Supreme Commander, and I've not even opened Neverwinter Nights 2...

Now it is back to work. I am still around here for the next three months, after which...?

Done, will submit when the office opens in the new year. yay!

I don't quite believe it.

For the last week, I have had one of theeese. The Eee PC, a tiny subnotebook. 4GB solid state storage, runs Linux, has USB ports and a decent VGA output. Most of what I'd want from a take-anywhere machine other than long battery life. Impressed so far.

With little thanks to CityLink, who recently appear to think that attempting to deliver packages is a bit much to ask.

Not really had time to play with it much, of course, as I am insanely busy working on my thesis. The only other thing I've had time for (which is to say made time for) is running; last week was my first to hit the 50-mile mark. Well, I manage to eat and sleep too. Not much choice on that front.

Less than a month to go!

On Friday I made a declaration of intent to submit my thesis. This has to happen two months before submission, and involves signing some forms and giving an Abstract. Here it is:

Game semantics for object-oriented languages

This thesis investigates the relationship between object-oriented programming languages and game models of computation. These are intuitively well matched: an object encapsulates some internal state and presents some behaviour to the world via its publicly visible methods, while a strategy for some game represents the possible interactions of a program with its environment.

We work with a simple and well-understood game model. Rather than tailoring our model to match some existing programming language, we view the simplicity of our semantic setting as a virtue, and try to find the appropriate language corresponding to the model.

We define a class-based, stateful object-oriented language, and give a heap-based operational semantics and an interpretation in our game model. At the heart of this interpretation lies a novel semantic treatment of the phenomenon of data abstraction. The model closely guides the design of our language, which enjoys an intermediate level of expressivity between that of first-order and general higher-order store.

The agreement between the operational and game interpretations is verified by a soundness proof. This involves the development of specialised techniques and a detailed analysis of the relationship between the concrete and abstract views. We also show that definability and full abstraction hold at certain types of arbitrary rank, but are problematic at other types.

We conclude by briefly discussing an extended language with a control operator, along with other extensions leading to a possible core for a more realistic programming language.

Just out of the shower after my morning run. The weather is so different from yesterday, colder, grey and starting to rain where yesterday was overwhelmingly sunny, not a cloud in the sky uplifting morning. Whatever, I feel good today too, positive effect of switching to running in the morning. The downside is that it feels a bit harder, particularly the morning after a gym night (mind you the gym is harder after a morning run than an immediately preceding one too). Could be something to do with stepping up to 3 gym visits a week, with a 5000m row; I never was any good at rowing, but I seem to be getting the hang of it, mostly.

The reason for the morning running (other than a couple of nice sunny mornings) is that I'm stepping up work on my thesis. Saves time and gets the day off to a well. Still lots to do, going well at the moment but currently looking like I will finish in December. So if you don't hear from me, that's why...

Back to the running, we're now 3 weeks into the 2nd half of the year, and my logs tell me in the first half I ran 944km (586 miles) in an inconsistent fashion (some longer weeks and off weeks). Minimal goal for the rest of the year is 1000km, or better to hit 2000 for the year, with a gradual build-up and more consistency. Anything else is a bonus.

I went to the Pentlands with dearg last weekend. Nice walk, as the weather turned out to be decent. Met people I knew there again, Ben from my office who I haven't seen in some time, somebody else from my department, and my uncle George. I don't know how often I should expect to see somebody I know in the Pentlands; it is close by, and I suppose I know quite a few people in Edinburgh, so even if I'm only there every few months (and always on a passable-weather weekend-day afternoon) perhaps I should not be surprised to bump into somebody every other time, say.

I am curious how pastimes are correlated with profession etc. For example, it seems that computer scientists are more likely to be into hillwalking than J. Random Officeworker, but perhaps my one-sided and anecdotal evidence is entirely misleading. I have no idea what the true figures are.

I've returned to running again. Just a couple of weeks so far, but I'm enjoying it. For motivation and something to work towards, I've signed up for the Glasgow half-marathon (the Great Scottish Run) at the start of September. Hopefully I will be able to gradually build up my weekly mileage between now and then to be able to put in a good first effort.

Before that, I will be going to Seattle in August for FLoC (Federated Logic Conference). My (joint) paper was accepted for the GaLoP workshop (Games for Logic and Programming Languages); I shall also be attending LICS (IEEE Symposium on Logic in Computer Science).

I sent off that paper this afternoon. Feels good! Can finally relax, was hard work.

Going off walking for a couple of days. Bye for now.

Attempting to write a paper at the moment. So neither much time to be writing here nor doing anything to write about.

Still not running. Feeling lazy even though I do get some exercise. Think I really feel the need to do much more, should start in a couple of weeks once this work's out of the way. Not sure whether I will concentrate on running this summer or the big hill-days I fancy.

Argh. There is drilling. I shall attempt to work via ironic pseudo-angry shouty music.

Well, I haven't been writing so much this month...

I have an icky face at the moment. Horrible skin problem; an infection which I am taking antibiotics for. I've been on those since Monday, and it seems to be clearing up, although you perhaps wouldn't think that if you didn't see me during the week. I'd post pictures, but it doesn't look very pretty (I have this skin problem, see...).

Unfortunately I also have a problem sleeping at night, largely because of this.

In other news. I have a second year progress report for my PhD to hand in by the end of the month. Well, strictly the due date is in the past. A couple of weeks... So if I'm not very communicative, that could be the reason. Or if I'm writing entries daily, it could also be the reason. Perhaps I will find a use for that unproductive awake-at-night time.

Tutorials have started again. I get to have fun with Algorithms!

Also, I'm doing a winter skills course in the Cairngorms in a few weeks. Hope the weather is good to me... And my feet recover/don't blister...

I think that's it for the week. Should hopefully have a working phone/net line on Monday, the BT person is visiting. Fingers crossed.

I don't think I've ever really explained what I'm doing my Ph.D. on, in laymans terms. Let's try; please have patience and excuse the length, but it is hard to write accurately but non-technically. I assume you know that computers run programs written by fallible humans.

I am working towards an object-oriented programming language based on game semantics.

Semantics

Let's start with semantics. When I say "semantics" I mean "programming language semantics"; given a computer program, what is its meaning? The aim here is to give an formal definition which unambiguously assigns a meaning to every program one could write in some programming language. This is in contrast to the "traditional" methodology of writing stacks of unreadable technical prose, or the more ad-hoc method of defining the meaning of programs by what a particular computer (and compiler, operating system, etc.) actually does with them.

There are a multitude of ways of doing this, and I shall describe two broad camps. One can give a set of rules specifying how the state of some (hypothetical) computer should behave in order to run your program (Operational Semantics). Alternatively, one can give a mathematical account of semantics, where the meaning of programs are given in some mathematical structure (Denotational Sematics).

Diving straight into that second idea, we can think of a computer program as something that takes some inputs and gives some outputs. The bells and whistles of modern graphical environments may obscure this notion, but fundamentally you are supplying input to a program somehow, and it is giving you some feedback. One can think of this as a mathematical function from inputs to outputs, and this is the basis for much work in denotational semantics. Of course in actual fact things are not so simple, but there is a great deal of work in considering program semantics in the setting where data is represented by some structured sets, and computation is represented by structure-preserving functions between these (this is the essence of Domain Theory, as I see it).

Game Semantics

Okay, now we get to Game Semantics. When I say "game" I am thinking of a two-player game, where each party plays alternately, such as in Chess. There is some set of moves one can make, and in particular at each position throughout a play of the game, there will be some moves that one is allowed to make. A strategy for such a game is just a sort of formalisation of the thinking of a player~imagine that, before starting to play, you have decided how to play at any given position of the board (the practicality of this is not an issue for us).

The words Game Theory may be familiar to you, but are not entirely relevant to game semantics. Typically in game theory one talks about players choosing strategies in order to maximise some outcome (get rich or whatever), possibly involving some "dice-rolling" in the process. The games I'm talking about do not have a notion of winning or losing; that's not what they're for.

Instead of thinking of a computer program as some function on structured sets, one can think of it as a strategy for some particular two-player game. I find this hard to explain, so let me give an example. Imagine that the program in question is the suitably simple "doubling" program: given some number as input, it will give twice that number as output. Okay, we could just think of it as a function on the natural numbers, but bear with me. I want to know the result of the program, so I ask the program "what is your value?". The program responds by asking me the value of the input; I proceed to say that the input is 21. The program then replies that the answer is 42.

It is impossible to decide the worth of game semantics on such a trivial example. The field has had some successes, but is probably not the answer to life, the universe and everything. In any case, there are particular types of games and constructions on games which we have been investigating (and now we're finally getting to what I'm actually doing). I have a developing picture of some mathematical structures, which will hopefully support some particular programming language, which will be designed to exploit those structures.

Object-oriented Programming

Okay, here's the other half. I mentioned an object-oriented programming language. Perhaps this is less important, and this may not be what I want in the end, but we will stick with this for now. The fundamental thing in an object-oriented programming language is, funnily enough, the object. Based upon the real-world analogy of objects which interact with each other in various ways, we view a program as composed of a bunch of objects, where each object may consist of some data (for example the entry for a person in a telephone directory) and some ways to interact with that data (continuing the example, one could ask for the number of a particular person, or update their address).

My Research

I don't think the OOP paradigm has a great particular worth, but it is currently popular, and no doubt attempting to address some of the nicer and more familiar features in a principled way is a good idea. It seems that one could create a reasonable language with object-oriented and functional parts based on the aforementioned games, and it would be nice if that was an outcome of this research. Crucially, I am not trying to design a realistic programming language; my research is all about some semantic ideas, and any language which emerges will be a way of investigating those ideas rather than an end in itself.

Maybe next week we'll talk about Category Theory.

Okay, so I don't spend all my time watching films and reading books, despite the preponderance of such in this weblog. I just spend most of my time doing things I can't talk about here. Top secret! Well, no, just generally incomprehensible or boring.

Research-wise, I'm alternating between getting not much done and being productive and excited about my ideas, in quanta of about a week. And everything else seems to go along with this; when I'm doing good research, I'm reading lots of books and watching movies, and when I'm not I'm... not. Of course this correlation says nothing about the direction of the causality. Perhaps it's something else, I don't know.

Anyway, just now I'm on a research upswing, and I'd like to be able to write about it here, but it wouldn't really mean anything. Well, at the moment I'm in a state of knowing how much I don't know, but even before this current foray into the unknown I could barely describe what I'm doing to the people in my department, never mind a general computer science community or the world.

In other news. Particularly enjoying climbing wall recently. On Sunday I was having fun with overhangs. I finally have enough strength to make some progress there...

Apparently my comment system is broken, and I can't edit my last entry. Bah. Perhaps I shouldn't write code. Okay, until I fix that, note that Kontroll featured a bear girl, it was not a bunny suit.

Currently got a thesis proposal to write. Head down.

To summarise recent events, the flat is now sorted and properly furnished, i.e. I have a computer desk and chair here, and there are bookcases. Thanks muchly to the mass of people who seemed to end up helping in this regard. I seem to be getting on well PhD-wise, as much as I can tell, and it's a very worky time at the moment.

It fluctuates. Yesterday I had a great day, in uni for quite a long time and working well. I found my reading interesting and engaging, and I think I understand much more than I did before. Why do I occasionally have a day like this, and then have days where I'm lucky if I can concentrate for an hour? It'd be good if I could work out any factors, I'm fed up with the other sort, the catching-myself-staring-into-space type of day.

These good days make me glad I'm doing a Ph.D. Although I didn't seem to have much time to relax this evening... Hopefully, I can get enough to start a new book soon, although I'm more worried about getting things done this week.

I feel like I'm progressing PhD-wise; certainly I'm finding game semantics and related ideas interesting, and in particular I'm finding it motivates me to learn the appropriate bits of alternative logics, category theory, and so on. Hopefully my thesis will be in this area; conversely I hope I get more interested in the alternate topic of computational effects (semantics of, a particular program logic for).

At least I'm currently motivated to learn the bits of category theory which tie these two things together, in a loose sense.

Please excuse the silence.

What happened? Well, after waiting in Edinburgh for a mail-order package, I zipped off to mum's for Christmas. Well, I say zipped, I seem to get all the late trains these days. That aside, I offed there for xmas with the family, fairly busy table for xmas dinner this year, even the garden table got in on the action.

Nice seeing people, relaxing a bit and not thinking about things. Note to self: get hold of Metropolis (1927) some time, although it's unfortunately not as cheap as the marvelous unrelated Metropolis (2001) (Anime). Read some more of my current novel (on book two of A Scots Quair), which I think I'll continue tonight, even though the mucker's lost its bookmark.

A bit after xmas, it was back to Edinburgh, where Liz+Adam and then Andrew were coming. Good to see them again, we had a fairly laid-back time, including staying in for Hogmanay watching movies. Which was just as well given the cancellation. Apart from the less relaxing shopping we didn't do anything too strenuous, apart from the enjoyable walk up Arthur's Seat with Andrew and Adam. After getting to the top and back, Adam (4) was still eager to go visit the lake and watch the birds, though by that time the day and clouds were drawing in. I look forward to some more walking with both of them.

Just starting to get back into uni. On this front, my paper for Trends in Functional Programming was accepted (yay!), so I should be making a couple of changes to that and returning it in the next couple of days. Lots of other things to be doing, many things to be reading about, and soon some taught courses to attend (though not overly many). Algorithmic game theory looks interesting, as well as some of the courses which will run later in the term. Enough to be busy with...

I'm sure I missed 1,000,001 things there, so I'll be seeing you soon.

Suddenly it's December...

Just come back from Sainsbury's, laden with exotic fruits and cheese. I now have dolchelata, blue stilton, applewood smoked chedder, german black-pepper coated salami, mango, paw-paw, sharon fruit and a canteline melon, among other things. Oh, and scotch bonnet chillis, which look funky. Should be a nice week on the gastronomic front...

Active uni stuff sort of finishing for the term, giving my last lecture thingy tomorrow on intuitionist logic (mainly semantics and the Curry-Howard correspondence I guess). Means I get to do reading on interesting things now I have time. My supervisor told me a bit more about game semantics yesterday, which was really quite interesting, so I'm looking forward to things. Reading up on that, category theory and type theory, maybe some more concrete language design things too. Should be fun.

I wonder what audience I'm writing for.

Tell me I'm not going to have only one long useful day a week.

Staring at a screen blankly...

I wonder if it's because there's too many people around. Not quiet enough. Bet if it was night time in JCMB I would find it easier.

Just waffled for an hour about recursive functions and models of computation.

Surprised that I hit that length fairly accurately.

Just matriculated. About 3/4 hour. Read some of my current book, From the Land of Green Ghosts. Worryingly the card is valid until 2007...

Food

Well, the Thai thing turned out to be rather extravagent. A great deal of fresh ingredients and various dried herbs bought. Thursday was a nutty chicken and pepper dish (slightly let down by my lack of rice noodles, and perhaps not a strong enough nutty flavour). Unfortunately I had to chuck the first batch of rice I made, and rather forgetfully failed to wash then overcooked.

Friday was a green chilli chicken from green curry paste lovingly made by my own two hands, followed by a mango parfait.

Saturday was a sandwich and reheated leftovers :). Not sure what I'm doing tonight, something with chicken.

Study

Now, on Friday afternoon I had a meeting with my Ph.D. Supervisor, feel better for having done so. I have a list of the taught theory postgraduate courses that are running this term, and the knowledge that I need to pay particular attention (as in mucho further reading) to category theory. I have a book and need to borrow books from the library.

Nice term starting again, feels somehow comforting. Oh, and I'm looking forward to learning things, of course.

Shiny Things

In a hopefully unrelated note, aquired Warcraft III off my flatmate yesterday. He bought it on Friday but it was crashing on his computer. So now I've completed the first (human) campaign. It's okay, spiffy graphics of course. As always there is a tradeoff between games that look good and have depth to them, but it doesn't fare too badly in comparison to Warcraft II. The system of hero units is interesting, and I think I like that. Basically it's a lot of fun, but I also want something in the empire-building arena, it's always a shame to leave your painfully constructed base behind.

Been playing Vice City quite a bit recently. Is fun. Great soundtrack. Is great game. Must... resist....

Wow. Didn't really realise how close term time is, it's just sunk in.

Next week is freshers' week, I have postgraduate "induction" stuff (lots of which can hopefully be skipped, seems dull), matriculate etc. Then the week after that, I start my Ph.D. Woo. Actually doing it.

And finishing work. Few final features to add to the compiler, do some polishing, cleanup and documentation. Oh, and I'd better rewrite the paper at some point so I can submit for refereeing, perhaps I should do that next week before term gets underway.

Ceremony happened.

Wandering around with my robes on. Waiting. Graduating. Tapped on the head with a hat of much esteem.

Reception. Lots of friends and course-mates. Prizes awarded: I am best BSc. Computer Science person and best Computer Science graduate (the latter from Sun, who refused me an internship last year... Heh.). Fun. Many congratulations. Beaming.

Thanks everybody. It was a great degree, everybody's been great supporting me, I'm so happy to have done so well with it. Cheers :)

Getting tiresome. Ceremony.

Not to brag, but...

Okay, I guess to brag. Apparently I'm the best Computer Science graduate from Edinburgh University this year. See their page for details.

A huge thank you to everybody who has congratulated me on my degree result (got some cards this morning too!), much appreciated. You're all wonderful people.

Degree classes are up. Result!

My degree is now complete. I have presented my final presentation (went okay) and all work is done. Now I just have to wait for my results. Come on!

Starting work soon. Be good to get into it. I'll get to read lots of papers to understand things :) And implementation, woo.

Oh, I now have an unconditional PhD offer.

Completion of undergraduate studies looms.

Final project presentation tomorrow. Then free.

Maybe some EdLUG tonight. Visit Liz for weekend. Start work soon presumably.

I'm quite confused. Surely I must have something urgent to be doing?

So my presentation looks like being next Friday. I'll have to do some preperation for that, slides and work out a demo, but that should be fine. I might find time to integrate my project work with the current compiler, which would not only be useful but would be something nice to say during the presentation. Shouldn't take too long, as the major changes are at a higher level. Hopefully I'll find what changes I clobbered from my CVS too.

May find some time to hack on the back-end for here, maybe even create a preliminary thing to distribute. Major issues at least include moving to templating. Then I might get to play with it elsewhere.

Hmm. Considering what it would take to do a non-db back-end. Oh and the fried/half-baked/baked continuum.

Three bound copies plus two abstracts to the ITO.

Again. Oops.

That is it.

Three bound copies plus two abstracts to the ITO.

Just the presentation to do, and that's it. Over. Yay.

Draft of report handed to supervisor today. Needs much polish, but it shouldn't be too lacking. Final bound copies for noon Wednesday.

Positive noises regarding PhD funding etc. Reassuring.

I'm handing in a draft of my project report (dissertation?) on Thursday. This is worrying. The final date is noon next Wednesday...

Theoretically I'm well set up, but I don't seem to be making much progress on the remaining stuff. Naturally I'm worrying more than working.

I don't even seem to have 40 pages of actual text. It should probably be about 50 pages (40-60) without appendies...

Hopefully I'll actually work late tonight. Time to invest in shares in chocolate/coffee/energy drinks vendors.

Bah.

After teasing with my abstract, I now aim to utterly confuse the casual reader. Perhaps if you persevere I might make sense at some point. On your own head be it.

Here is an example in the language Camelot (see abstract), computing the factorial function. For the non-mathematician, n factorial (written n!) is n*(n-1)*...*2*1.

  let rec fac n = if n = 0 then 1 else n * fac (n-1)

This trivial code snippet can be found in any introduction to functional programming, and indeed in many of my project talks this is also used. One quickly grows weary of it, but I will persist. The next step, as any fule kno, is to reformulate this to be in a "tail-recursive" form. There is little point in me trying to explain this properly, but suffice to say that this form permits optimisation.

  let rec fac' n r =
    if n = 0 then r else fac' (n-1) (n*r)
  and fac n = fac' n 1

This longer code should be much faster. That's how optimisation works. Notice that this time no computation happens in fac' after a function call to fac' - the only call is fac' (n-1) (n*r). This means that the result of the function is the same as the result of that function call, which means that nothing needs to be remembered along the way.

Anyway, you probably didn't understand that, but I shall press on regardless. The important part, fac', becomes the following when compiled to Grail:

   method public static int fac' (int n, int r) =
   let
      fun f:fac'(int n, int r, int t:8, int t:9) =
         if n = 0
         then r
         else f:0(n, r, t:8, t:9)

      fun f:0(int n, int r, int t:8, int t:9) =
      let
         val t:8 = sub n 1
         val t:9 = mul n r
      in
         invokestatic <int Fac2.fac' (int, int)> (t:8, t:9)
      end
   in
      f:fac'(n, r, t:8, t:9)
   end

• It's much bigger. Grail has a much restricted structure, and each particular piece of syntax does less than something similar in Camelot. This is the basis of compilation.
• It looks similar. This example doesn't show it as much as many would, but notice that the syntax of if/then/else is similar. This is because Grail is a functional intermediate language.

See the part that says invokestatic. That corresponds to the call of fac' in the original. A method invocation is relatively costly, as it takes time and memory, but luckily we can eliminate this because of the tail-recursive form of the original. The first part of my project does this, giving:

   method public static int fac' (int n, int r) =
   let
      fun f:fac'(int n, int r, int t:8, int t:9) =
         if n = 0
         then r
         else f:0(n, r, t:8, t:9)

      fun f:0(int n, int r, int t:8, int t:9) =
      let
         val t:8 = sub n 1
         val t:9 = mul n r
         val t:10 = t:8
         val t:11 = t:9
         val n = t:10
         val r = t:11
      in
         f:fac'(n, r, t:8, t:9)
      end
   in
      f:fac'(n, r, t:8, t:9)
   end

Extra variables are generated here but not actually needed. They are generated because in some cases they are needed (to swap the actual arguments so that they match the formal arguments syntactically). Anyway, this gives a good demonstration of the second part of my project, which is the elimination of local variables. :)

   method public static int fac' (int n, int r) =
   let
      fun f:fac'(int n, int r) =
         if n = 0
         then r
         else f:0(n, r)

      fun f:0(int n, int r) =
      let
         val v2 = sub n 1
         val r = mul n r
         val n = v2
      in
         f:fac'(n, r)
      end
   in
      f:fac'(n, r)
   end

Now instead of n, r, t:8, t:9, t:10 and t:11, we only have n, r, and v2. This is good! While t:10 and t:11 could have been avoided in this case if care was taken in the earlier optimisation, it requires the analysis of the latter to see that t:8 and t:9 can be done away with. Plus, it looks better, there are less things between the parentheses. ;)

Now, what you saw was and wasn't the meat of my project. Those were the effects of the 700 lines of ML code I wrote, yes. But that really wasn't much code (there are probably only a couple of hundred lines of real code).

Most of my time was actually spent analysing what goes on, proving formal statements about the optimisations. How much memory they save (or don't!), and so on.

Recently I have been writing more substantial examples, as there were none, and running them through the compiler, to show that there is a real point in these optimisations. It seems there is. That was nothing other than a complete triviality which I just demonstrated. My first substantial example involves priority queues; 19 functions are implemented in 119 lines of Camelot, which becomes 703 lines of Grail...

I'm sure it would only be about 20 lines of ML.

If anybody is in a place to pick holes in or ask intricate questions about that, then would you like to proof-read? ;)

And I had to modify the back-end of the diary system because this was so long.

For those who have been floating along bemused by talk of my CS4 project, let me confuse you further. Here is the first draft of my abstract:

Abstract

This project studies the optimisation of the compilation of Camelot. Camelot is an experimental functional language designed for resource bounded computation, and so we study optimisations improving space usage, and the effects of these optimisations on provable resource bounds.

Tail-recursion optimisation and local variable consolidation are implemented and studied, with a detailed correctness proof for the former, and characterisation of the effects of both on stack usage.

Practical significance is shown via a set of non-trivial example programs.

Doesn't that make you feel all warm and fluffy inside?

I'll spare you gritty details for now.

Well and truly restarted project work now.

Sorry to be boring and write about Uni, but that's been my main concern for a while. In other news, I can reveal that Liz+Adam+Rob are up visiting this weekend, which is nice. It's been a while. Finally saw my dad recently, which was good. Um. What else. Bought a cheap book today?

Now back to uni. Finally registered to graduate now that my exams are done.

And my project report. Actually a dissertation. I seem to have 30 odd pages (probably aiming at 50), but a chunk of that may need taking out into an appendix and the rest needs reworked. After a meeting with my supervisor on Thursday I was reinvigorated, but also rather worried at my progress (even though it's good compared to others). Now after starting to put work in I feel better.

I may even get to do some more implementation, providing a break from all the writeup. Hopefully if I decide to, it'll be straightforward, but otherwise at least it will motivate me to continue writeup rather than get into other distractions.

Oops. Left this submission lying over-night. Just about to go off for a wander to Forbidden Planet, then some work.

By this time tomorrow, I shall be free of exams, free!

Please?

Supposedly revising graphics. Can't do more than skim. Too much material. Yargh.

Security should be okay.

Woo!

Just had my fourth exam of six. FPLS, and it was an easy exam. Woo! Two straightforward questions, and one based on amusing properties of a mathematical structure called a CPO, which AFAIK nobody did. I'm sure it was more fun to write though. If the rest of my exams were straightforward and easy I would maybe have tried it, but as it is I'd like the marks thanks.

It is sometimes hard to contain my desire to tackle harder questions to prove I can do them, rather than feeling like I'm taking the easy route, but in this case the distinction was...well, very distinct, and I didn't have so much trouble.

As for the others this week, of which I have not written, Computer Algebra was naturally harder, and I struggled, though it was probably an average paper. COC was a nice paper, but unfortunately I was actually ill, so I dropped some marks. Lots of it was possible to do without thinking, but when it came to working out what to do an induction on my brain deserted me. Luckily not worth many marks.

So I think I'm doing okay so far. 2 to go.

As I don't think I mentioned it, my first exam was on Tuesday. It went okay; I wasn't completely perplexed, but I had a terrible trouble choosing which question to do and probably made the wrong choice. Still, I was able to give some semblance of answer to each part of the questions, some of which will be more reasonable than others, so I should get a decent mark. To possibly enlighten you, our exams are 1.5 hours long, choose 2 of 3 questions.

Anyway, I'm mainly revising Computer Algebra just now, which I find to be a relatively hard course. It's the middle of the three exams I have next Tuesday, Wednesday, Thursday.

Unfortunately I'm finding that I became less productive since I brought my sleep pattern back into synchronisation with normality. I find it much easier to concentrate in the dead of night. Still, doing okay. Here we go.

My first exam is the day after tomorrow. This is stressful, and I'm really having to devote all my time to revision just now.

Sorry. Even short interruptions can kill my train of thought for quite some time, and put me off for hours. Short diversions, of course, are necessary, but are only positive because they're intentional. I should surface again in about two weeks.

In which I waste time by cataloging the time I have not wasted and analysing this with respect to expected productivity and psychology. Sorry about the length, I have been posting a bit recently...

Resolution: I shall not rewrite my revision timetable. Too often.

Much of my revision has involved making notes, on the principle that writing things down gets it into the skull, and I can keep track of where I am. I have been trying to only make notes once I understand things rather than unproductively copying things. Here, ordered by date of first exam, are the numbers of sides of A4 paper used in my brief notes:

• Computational Complexity: 4
• Communication and Concurrency: 2.5
• Computer Algebra: 3.5
• Formal Programming Language Semantics: 1
• Computer Security: 2.5
• Computer Graphics: 15

Two things are obvious here. Firstly, there are too many Computer Science modules with "Computer" in the title. Secondly, there's an obvious variation going on there. But why? Here are some possible explainations:

1. Courses differ in volume of content and difficulty of material
2. My perception of relative volumes of content differs
3. My ability differs
4. I have covered material during the year to differing degrees
5. I prefer to revise particular subjects
6. I'm complacent about some subjects

Now, there is probably an element of all of these. I'm convinced that some courses are just easier, but there's a judgement/abilty call that's hard to make. And oddly enough, I hadn't covered Graphics much (basically what I had to know to draw a teapot, hence in-depth notes, but I genuinely think that would be hard to summarise breifly without just giving a "topics covered" list. Perhaps due to range of topics.

Interestingly, I also didn't bother with many security lectures, preferring to do other course/project work, but I haven't made as many notes. Is this course slower for wide audience appeal, or am I more apt? I have certainly come across some concepts before.

Seriously.

That's what comes of going to bed after revising graphics while thinking about forthcoming graphics revision. When I woke up, I was initially unable to determine if I had been asleep, or, for that matter, if I was actually awake, as the subject matter of my dreams and thoughts was so similar, not to mention so dull. Dull enough to put me to sleep...

Revision may or may not be going well; I feel both ways at different points. In graphics for example, I don't feel like I've got anything like the course covered, which makes me all negative, but at other times I remember I'm progressing day by day, and still have time left, so I should be fine. Time will tell, but currently I'm positive.

Now, curry.

Become nocturnal. Oops.

But it was a beautiful morning this morning. So much so that I hillclimbed after breakfast. Nice.

Good revision progress. Actually understanding some Computer Algebra, Gröbner bases and whatnot.

Got some marks back today, including Graphics finally. I'm pleased. Also looking at the percentages of coursework/exam for different modules, it's not quite what I'd expected.

Shopping. Good. Freezer decided to defrost itself. Bad.

Now it's half five, maybe I can get down to the meat of the day. Becoming slightly nocturnal.

Just recieved a letter through the post formally offering a PhD place (conditional on funding). Of course there is yet another level of formal approval which is actually a formality, but I have it.

Yay!

Now hopefully I'll be funded too.

No hillwalking today.

Doing formal programming language semantics today. The reason for the quick changearound is that this is my preliminary week, surveying what I've got to do. I've been spending one day per course (that takes me up until tomorrow), then I'm going for more sustained bursts. I seem to try a different strategy every time; I'm not sure of the merit of this approach, but to be fair it hasn't served me too badly up until now.

Politicions seem surprised that there's guerilla warfare in Iraq. Hardly surprising. C4 news ran footage from before the current conflict started, of Rumsfield and various .us politicians. Interesting how they basically said it'd be a walk in the park. I'm more inlcined to think that they knew exactly what mistruths they were peddling, and believed it wouldn't matter once war started.

US making noises about Syria arming Iraq. I wonder if it's time for us to head to the hills; do you think Bush knows who armed Iraq before?

Amusing to hear the Syrian spokesperson respond generally to the concept of war on Iraq rather than the specifics of arming them.

Good morning this morning. Went for a little walk up arthur's seat. I should do that more often.

Now I'm off for a complex day. (Revising complexity)

Troops encouraging an uprising in Basra, precision bombs hitting their targets. Changing times? But yet more friendly fire, bombed market... Bush not convinced of the role of the UN in post-war Iraq. Har de har har.

Revising teapots today, apparently.

And now for something completely different.

A day over 3 weeks until my first exam. 31 days of revision between now and my final exam, including time between exams. Just under 4 days per course. I can do this.

Since I last wrote anything here, I've had an interview for a PhD place. Nothing official yet, and money is another hurdle, but from informal feedback it's looking good. Also (and more concretely) should be working at LFCS this summer (LFCS is where I was last year).

Just writing a revision timetable. Where are my highlighter pens!? All relavent Red Dwarf jokes have been made at this point.

Actually won a couple of games of pool last night... Some bizarre games though. Wish Teviot would be open for pool over the holidays.

Work handed in. So, this is the end of my last non-exam term of my undergraduate career. Scary. Weird. Huh.

I'm probably going to have an interview for a PhD place at Edinburgh next week. Woo.

Exams in just over 4 weeks. Project. Um.

Liz comes tonight. Woo.

Ostriches and Emus.

SMS spam :(

Mmm. Pool. I play better than I used to, but I've a knack for losing the endgame.

Complexity going fine. This assignment seems too routine. Is this a relaxing final piece, or am I missing a vital subtlety? Tune in next week.

Finals in five weeks. Scary.

Grump. Last night it was pointed out to me that my picture gallery is suboptimal, as the thumbnails are tiny and the actual pictures are huge. So what I want is a page of thumbnails which links to medium sized pictures (with prev/next links) which link to the full size beasts.

I found one tool which would generate such a gallery, but the images it generates are broken. Hence the original pages are regenerated with the dodgy tool which works, and I'll try again later.

In other news, I had a nice day yesterday. Somehow I spent all day being mildy productive and not wasting time, and yet I don't have terribly much to show for it. Oh well, at least I have something and ideas on where to go from there.

It's much nicer now that I'm thinking of stuff I write as contributing towards my project report. I think if I finalise the structure I can get on with writing the expositionary sections, such as an initial chapter setting my work in the context of the MRG projcet, outlining the shape of Camelot compilation, and discussing the possibilities for and usefullness of potential optimisations. Actually that sounds like three chapters, which is fine.

I think I will have to watch the length, it has to be 40-60 pages, and that is really not much. I'll need to edit the introductory parts to be comprehensive without waffling too much, but I'm better off waffling initially, because otherwise I'll never get started.

Exams are going to start looming. I'll have to work out details, and plan my revision. And so another year comes around.

I had my project meeting today, and gave my talk as expected. I actually scraped together the organisation to make printed transparencies, which were pretty and nice because I used our Informatics slides document class. I think I said what I intended to say, though as always I'm not so great at presentations. I think I'm improving. Now if I had a talk which wasn't split between progress report and content, and was long enough to explain things properly, I could do okay.

Must do computer algebra work tonight. And buy some food, and things.

And possibly even relax.

Now to work on my project. A presentation to give to my project group on Friday (will the calibre exceed that of my previous talks?). And code to write, so that there is some content to my project. Oh, and I should be working on the report.

Going okay on my Computer Algebra assignment. Actually got work done on the train, and understand things. Good progress, and that's a first. Just fetched the Complexity sheet too, must start looking at that and get it mulling over.

Wonder if I can sustain a late night tonight? Get some coding? Time will tell. A decent computer would help enormously, manufacturer still hasn't got back to me on my Laptop.

Missed a meal last night. Oops. Completely forgot. Got a phone call on the train asking where I was. I said "approximately York, on a train". Subsequently I found out that Edinburgh would have been more appropriate. Here I make a note in my diary: "consult diary more often".

I've been meaning to sort out what I'll do when I graduate for quite some time now. I really don't feel like getting a stressful job, although the money would be nice as always. Really, I want to continue to study. And yet, I had not made any applications, and time is getting on. I do procrastinate so.

Yesterday, my CS4 project supervisor suggested I apply for a PhD here at Edinburgh. While the Compiling Techniques lecturer suggested that last year, it's theory that I really want to do, and it's nice to know that somebody in that field thinks I should apply. So I'll be filling in application forms soon, much fun as it will be. And considering carefully the direction I want to go in towards my thesis.

Oh, and he also wondered if I would like a studentship working on MRG again this Summer (no, I did not have a job lined up).

I'm quite happy and optimistic today.

More project work. And I'm starting to use LaTeX a little more to my advantage. I wrote my first theorem today. And some lemmas. Whee!

Working some more on the compiler. What fun. Actually it's nice to touch code again.

Need to think hard about where to go from here.

Fun-filled weekend of work ahead of me. Hopefully I'll get a large chunk of non-project-work out of the way. I have to, I'm going to be at Liz's next weekend.

My father comes back from New Zealand on Sunday, I think. I'm not envious in the slightest, honest.

Well I did take some time out and go to juggling. Wasn't really in a juggling mood, but I played with a Diablo for a bit, got shown a few basic tricks. Unfortunately, owing somebody a pint, I had to go to the pub afterwards. The swift pint turned into about five hours at the pub, including playing pool and lying around on pool tables. I blame the company. Jugglers.

Project work going okay, trying to write up a short document before I can look at non-project things a little. I may get to do something more exciting for a while, implementation of a novel technique, which would be hard but maybe fun.

I just had the opportunity to exclaim these immortal words.

(The door was locked.)

Muchos fun doing project work just now. Writing up things related to my original idea rather than the horrendus side-track of correctness, but I'm unfortunately getting a little handwavy. Still, I'm doing something, and will have something to show on Thursday.

Hopefully I'll get to expand my optimisation to be less stupid and trivial, getting to write some actual code, before having to do more reasoning again.

Might or might not go juggling tonight. I've stopped going for some reason (i.e. workload), but I'd like to restart even if I do have no time.

Long time no wibble.

So, I went down to Liz's the day after my last entry. Amusingly, all three of my trains were delayed, by varying times. Fortunately, the first two delays were well thought out. Rushing into Waverly with oh-no-help-I'm-going-to-be-late seconds to go, I saw that my train wasn't there yet, and got something to drink. Arriving at the platform 10 minutes after the departure time, the train wasn't there, and a delay was announced. 10 more minutes, I'm off. Great.

Arrive at first station. Oh dear, missed my connection. Wait, no, it's late too, so instead of a 1/2-hour wait I just need to walk to the next platform and get on. Great!

Getting into Meadowhall, on the other hand, I was less fortunate. Missing my connection, I had to wait about an hour for the next train, with such amusing interludes as an appropriate train being listed for quite some time before just disappearing from the board.

Anyway, to cut a... to finish the story, I arrived in Barnsley in the early afternoon, and Yea! there was much rejoicing.

Cut to Christmas eve. Going to bed at 2am, I discover Liz ill with a sickness bug caught off Adam. This lasted the better part of Christmas day as sickness, and afterwards as general illness. Just past the stroke of midnight on Christmas day, I caught the bug. So we had a fun Christmas. Really.

At least we managed a nice morning. Sitting opening Christmas presents with Adam was wonderful. He meticulously read all his new books before opening the next present. Wonderful to watch a 3-year-old's logic at work. It might be the first Christmas he remembers, too.

Slowly recovering, and starting to think of dull work-like things. Then meet! Whole bunch of afpers gathering at Liz's parents house for new year, along with the better part of the surrounding week. Lots of nice people, some of whom I had not met previously, and generally a happy time. New Year was better than Christmas.

Nearing the end of my stay; last night with Liz. And it snows, good and proper! The sky was nicely covered, the air full of heavy snowflakes floating down hypnotically. Lying in bed, watching the snow fall together... I shall remember this.

Do you notice that I get less verbose the further down the page I get?

Unfortunately I had to come back all too soon. More unfortunately, my laptop has given up all semblance of working order, and needs a good beating and/or repair person. So, having been unable to do the work I intended due to illness and lack of laptop, I came back to a pile of work. I seem to be making some progress now though.

Project status: stalled, but about to be attacked viciously. New courses: seem interesting. Naturally there is a direct relationship between attendance and waffle. I'm attending two "small" courses, circa 10 people taking each (Computational Complexity and Computer Algebra) and a "popular" course (Computer Security). Security shouldn't be too challenging and should be interesting if overly verbose and insufficiently detailed, whereas the other two should be both interesting and challenging. I wonder what I'm in for.