First, I had the waking dream, part of which concerned a theatre in north London which we used to visit from time to time, to see stuff like Ibsen or Chekhov. Neither a West End theatre like the Duke of York’s nor a National Treasure like the Old Vic. More like the Almeida, but not as far north as that, and not quite as awkward to get to from Epsom. A newish theatre, perhaps like the Thorndike theatre in Leatherhead (actually, I learn from Wikipedia this morning, a repurposed cinema. Not sure I ever knew that). Then, as I became more awake and got to thinking a bit harder, perhaps it was actually in south London. Or perhaps Goldhawk Road? The Caledonian Road? And now I am fully awake, I don’t think the place exists at all. Unusual, as things of this general sort in my dreams do usually have some kind of existence in the real world.
Second, I had the strange noise outside. First thought was that perhaps it was the wind in the trees. Went to the window, the top of which was open, and thought that maybe it was a train, perhaps a goods train rather than a passenger train (we do get some such at Epsom), on an early morning movement. A little later I heard the noise again, this time certainly not a train, more like a large aeroplane, quite a long way away. Eventually I decided it was some trick of the weather or the time of day and that what I was actually hearing was the hum of the M25, two or three miles away and not usually audible.
Then third, my thoughts reverted to the computer files that I had been thinking about the day before. How did their existence relate to that of psychic phenomena in the brain, sometimes postulated to be no more substantial than the fetching, two dimensional attractor states of a dynamical system? Nothing so solid and substantial as a bunch of neurons and synapses which you could, as it were, put your hand on.
So, on a regular computer, most of the information is organised in terms of files. An almost universal device for the structuring and management of the huge amount of information concerned. So there will be a file management and a lot of files, some of them regular files which regular users get to use, some of them system files, not always visible, let alone accessible, to the regular user. One of the system files which is particularly important is the file catalogue, the list of all the files in the system, perhaps the list of all the files that have ever been in or that have ever been planned for the computer, and one can have long and loud discussions in the pub about whether the list of files can itself be a file.
But my concern this morning is the regular files, the files that regular users use. More particularly with the data files, as opposed to files which contain, for example, programs or images, data files which are organised in terms of records, organised serially one after another, with a first record and a last record. Perhaps the file is about all the cars in the forecourt of my garage, with one record for each car. Sometimes these records will be fixed length and fixed format, sometimes they will be something more complicated, but what is important here is that they are the unit in which the user, or rather the program on behalf of the user, works. The program is connected to the file and processes one record after another; a program which might amount to little more than something like ‘for each record in file X do Y’, where Y is something big, important and complicated.
Then we think that this file, so real and straightforward to the program, might, under the hood as it were, be some more complicated object, perhaps even a phenomenon or a process rather than an object.
The program thinks of as a record, once it has been read from the file, as a chunk of memory. Starting by convention at byte one, bytes one to four are, for example, the person identifier; byte five is that person’s age in years on the 30th June last; byte six was their weight on that day, byte seven their blood sugar and so on. With some of this information being bound to the file, some of it being bound to individual records and some of it coded into the program. Whereas, in real life, on the chip, we may not have a chunk of memory at all. We may have a series of fragments of memory presented to the program in such a way that they look to it like a simple chunk of memory.
The program thinks of a file as a chunk of storage, persistent storage rather than ephemeral storage in the way of memory, in every way comparable to the tray of punched cards which in the middle of the last century was the external manifestation of a card file. A steel tray or drawer, full of punched cards, each card about six inches by three, each card containing up to eighty bytes of information punched into columns across the card. The tray was the file and the card was the record. The catalogue might tell the computer operator where this particular tray was kept.
But, just as with the memory on the chip, the real file might be something more complicated. It might be held in memory or it might be held on some more or less external medium like a magnetic disc. In either case the file is likely to be fragmented, a fragmentation which is hidden from the program by the file management system, a fragmentation occasioned by the availability of space at the time the file was first written. A large file might be represented by hundreds, if not thousands of such fragments, although things can get a bit slow if it really is thousands and housekeeping should be asked for a consolidation. A consolidation which will probably include updating the file’s entry in the catalogue.
The file may exist in many copies. There might be some master copy which lives deep inside some mountain in Wyoming, but there might be copies all over the world, for more convenient access by all the users all over the world. All this being looked after by the file management system.
Then again, the file might be something built on the fly from some larger database, just in time, to use the jargon of supply system engineers. What the program sees as a file is something temporary, perhaps drawing together information from a number of linked tables in a relational database, the sort of thing you might buy Microsoft’s SQL Server to look after for you. Which gives rise to the further complication of the possibility of the real data underlying the virtual data changing in the course of the latter’s life. Does one forbid such change? Does one ignore it? All this being looked after by the database management system, for which you could, and probably still can, pay hundreds of thousands of pounds.
So the bottom line is that a file on a computer is not real and substantial in the way of a tray of punched cards, never mind of a tree in the garden. It is perhaps more accurately thought of as a capability, a package of functions. You can open a file, you can ask for the first record or the next record, you can close a file – but you cannot hold one in your hand. Alternatively, you might think of the information content of particular files. This file, for example, was created by the General Register Office at Titchfield in Hampshire on the morning of 3rd June 2015 and contains the then extant record for all the deaths registered in England and Wales in the course of 2013.
Whether we need to know all this being another matter...
PS: there is a different sort of reality issue with the Duke of York’s, no longer an independent theatre with a life of its own, just one of the many branches of the Ambassador Theatre Group, one of the small number of combines which have gobbled up most of London’s theatre land.
No comments:
Post a Comment