Drab as a fool, aloof as a bard

Nineteen Eighty-Four (henceforth “1984”) is one of those books that seems to follow you around, like Alice in Wonderland. By that I mean that references to it keep cropping up around you, but more than that, it’s like the references to these works somehow encode some information which underlies some central truth to the universe. Follow the white rabbit, as Morpheus would say. In fact it was at the intersection of these two great threads that I resolved to read 1984. I was visiting a very dear friend, who I saw had started reading this book, which piqued my interest in it again, but at some point in my conversation with, she mentioned one of her favourite books, Alice in Wonderland of course, and I said something like, “Yes, it’s very important for [imaginative artists] like you to read that book.” (or words to that effect) and then said “I wonder if there is an equivalent that [responsible software engineers] like me should read?” Naturally, her reply was 1984.Buy Chlorpheniramine Buy Macrodantin Buy Dexfenfluramine Buy Chlorotrianisene Buy Skelaxin Buy Monopril Buy Carbamazepine Buy Vinblastine Buy Iproniazid Buy Multivitamins Buy Trandolapril Buy Senna Buy Nialamide Buy Cyclamate Buy Dihydrotachysterol Buy Nafcillin Buy Dextromethorphan Buy Estrogen Buy Pyrazinamide Buy Phenytoin Buy Mebanazine Buy Tocainide Buy Corticotropin Buy Butriptyline Buy Sulfonamides Buy Echinacea Buy Aztreonam Buy Milrinone Buy Adderall Buy Digoxin Buy Meclizine Buy Butalbital Buy Prednisone Buy Bricanyl Buy Promethazine Buy Methaqualone Buy Pemoline Buy Aciphex Buy Aerobid Buy Trihexyphenidyl Buy Avapro Buy Adalat Buy Lopid Buy Levoxyl Buy Nexium Buy Mucomyst Buy Chlorothiazide Buy Disopyramide Buy Isradipine Buy Aurothioglucose Buy Lunesta Buy Piperidolate Buy Phenolphthalein Buy Sucralfate Buy Levlen Buy Oxyphencyclimine Buy Meridia Buy Pindolol Buy Pargyline Buy Acyclovir Buy Guanabenz Buy Phenazopyridine Buy Fioricet Buy Simvastatin Buy Imdur Buy Bayer Buy Cilexetil Buy Mevacor Buy Ethanol Buy Paregoric Buy Toradol Buy Orphenadrine Buy Apomorphine Buy Pyrilamine Buy Bupropion Buy Diphenadione Buy Acetylcholine

The short answer is “no”, in case you are happy to just take my word for it and don’t have time to read my justification for this view. A slightly longer answer would be “Hotwire version 0,710 would be the best piece of software ever (until the next best piece of software ever comes out) if a finite number of bugs were fixed and a finite number of relatively simple features were added.” Many of the bugs and features missing are related to the usability of the user interface, which is to some extent a matter of taste, but there are usability improvements that can be made based on certain widely accepted metrics. Below, then, I give a mixture of a review, an infestation(?) of bug reports, and a design concept.

Just when you start to get surprised at some of the complicated pages that IE7 can render flawlessly, you bump into another almost trivial page which the browser laughably fails on. I think I need to add some more tags for categorising my blog posts, as I don’t want to spend longer than necessary answering the question “Is this IE bug the same as one I’ve already reported?” It appears this one is new, though, so I will write it up. I should say that the bug does not affect Firefox, and I only found it after noticing that IE7 did not have the styling I expected for a piece of popular third party code. Admittedly I can’t state exactly how popular the code is, but it was at least of significant enough size that I did not want to consider maintaining a fork. The good news is I managed to work around the bug in this case with a one line fix, but that is largely to do with the nature of the page, not by finding a universal solution.

Freedom is its own reward, but it shouldn’t be surprising that choosing Freedom leads to many unexpected positive side-effects. For instance, Miro is designed to make online video more democratic, allowing anyone to become a publisher and anyone to subscribe to any publisher, without central control, but by embracing this model, users of Miro, if they are like me, find that they have access to refreshing new creative works, that change how they view media. It’s like the jump from TV to PVR, and the jump from closed source to “open source” (or at least shareware) all in one go. That is why suddenly going without it because of an upgrade to a separate program (especially after paying for a greater download cap from your ISP) can feel jarring. If you have found that Miro crashes at the end of each video (or just errors tens of times at startup) then my findings might be relevant to you.

I can’t really think of a neater title for this, which may mean this post is not easily discoverable on the Web by people who would benefit from it. Worse, I suspect that this is the sort of trick that people don’t even go looking for because it never crosses their mind. Still, I have to document it, not least for my own future reference. For me the problem it solved was trying to avoid the six-legged freak console-based text editors that I found myself using while SSHed to a remote server at work. I could either type vi file.txt (which is quick to type but ultimately counter-productive) or I could find a desktop text editor and navigate to the remote location I was accessing over SSH (which would take longer but save time by about the first drag and drop). Although I had not heard of a system which had the benefits of both these options, I one day realised that there was no technical reason I couldn’t fire up my desktop text editor locally, by running a command on the remote machine.

Blogging time has come round again, and with it the opportunity to make up for the relatively narrow appeal of my previous post. I’m glad to see, though, that my previous post was appreciated by someone who matters, which has inspired me to consider possible enhancements to Hotwire when the latest version hits Debian testing. In the closing hours of this month, my focus is on MySQL and sed, as hinted at last time.

2007-12 for me is “Python appreciation month”. Going a whole month being appreciative of it wasn’t so difficult, and was the least I could do after realising how much I had under-estimated a program written in it, particularly as that program is going a long way towards fixing the very problems with the console that I’ve been was bemoaning. I suppose this post belongs with the last one, and I could be writing about MySQL or a sed joke instead, but it is New Year’s eve, so I’ll have to hope that the blogosphere appreciates the quality if not the quantity of these posts, while I go off to a party.

While I stand by everything I’ve said about the problems with the console environment, I have tried to be open-minded enough to acknowledge its benefits. That’s not particularly hard, as I often make use of some tool that is (currently) only available on the command line or get a job done faster because it is well suited to a text- / keyboard-based system. Using the console is also getting easier and more productive, though, thanks in part to Python, strangely enough, but that’s a matter for another post. This post is about what can be achieved using just a standard console and utilities.

It is becoming clearer to me, the more I force myself to work around and against the quirks and usability failings of the terminal environment, that the concept has reached a dead end in its development. There are, fortunately, some brave efforts out there to try to imagine what a shell could be, taking into account modern advances in technology and interface design, often without throwing away too many of the good features that make legacy shells so popular. Some examples of this new breed of applications are anthias, cuiterm, fish and hotwire.

A colleague at work gave me another challenge recently, or at least they showed me something that Perl can do, which as usual caused me to seek out a more elegant or efficient implementation in another language. A previous challenge had been to write a speedy script that found common rows between two text files, and I had resorted to Python for my implementation, just to make it difficult for myself. After a few rounds of optimisation we found that, when run once, the Perl version was faster, but when run 10 times in a row the Python version took the lead. In this case, though, the challenge was to write a one-liner which would take a string as input and reverse the letters in it. As I enjoy replacing my colleague’s unreadable Perl syntax with a handful of greps, seds, awks, trs and sorts piped into each other, I thought I would aim for a pure BASH solution to this problem. I would include the Perl code for comparison here, but I’m sure the Perl supporters have put it on posters or something to advertise what presumably they believe to be an advantage of Perl over BASH (but they will run out of things to put on those posters once someone gets round to writing mod_bash...).

Having looked in my Mystery Box of potential topics for blog posts, I’ve found a couple of things which are relatively short and don’t really make sense on their own. The first is an observation about key combinations on the command line, and the second is a joke which came to me recently and I wanted to publish before someone else came up with it independently. By my calculations, the first half of the post will be technical enough to justify writing this, and the second half will be funny enough to justify someone reading this, but not by much.

Setting up a new database, or adding a new user privilege to one, and even creating a new table in one is something that even database administrators do relatively rarely. After all, if the number of databases or tables is changing faster than you’re querying the data it contains, then you’re probably doing something wrong. It is only natural, therefore, that the commands for these “one off” actions are more easily remembered, through practice, than the SELECT, INSERT, UPDATE, DELETE commands that make up the bulk of a database’s activity. Even if one can remember the precise syntax for the commands, or one uses a graphical interface where syntax errors are impossible, there are still corner cases where the semantics of the database server can be counter-intuitive. Recently I bumped into two such cases.

I recently had cause to make use of a language I had enjoyed messing around with before, namely XSLT. Unfortunately that language still hasn’t reached the critical mass of having strong support in IDEs, active user communities, and clear documentation, for which it seems some of the blame lies with the overly-abstracting, purely academic viewpoint of some people in their ivory towers that no longer seem to be solving the real problems that exist. Instead of a coherent view for XML technologies forming and being adopted, there is a rush to generate more and more over-arching, under-supported, paper-tiger standards, with the community expected to bounce between them and fill in the gaps while finding extremely niche or unnecessary use cases to justify their existence. The experts even accept there are problems with the core of XML itself, but how many people do they think are making use of, or even know about, things like XSL-FO, XPath, XQuery, XLink, XPointer, OWL, RDF, SPARQL, Turtle, N-Triples, Notation 3, and SKOS? It is therefore not without reservation that I delve into this world with XSLT, but it did seem like the right tool for the job and a rather fun logic puzzle.