For the past three nights, the person in the room below me has been talking into the night. The floor between our rooms must be pretty thin as I can hear every word (although to be fair, I can't understand them). He stopped at 3:30am on Saturday, and 2:30am Monday. I'm starting to go mad. I plan on having a word today but I don't expect a great response.
I also plan on buying some ear plugs. I also added a date to this post.
Plans: newsdot clone, XML powered stuff instead of SQL (particularly links); wiki stuff... blended into a soup of scripts, xml dialects, translations... As Larry Wall put it, built not like a building designed to stand the test of time, but a film set, thrown up to serve a purpose, then ruthlessly butchered and chopped and changed to meet new demands.
I thought I'd try out ion simply as an exercise in open-mindedness, but now I find it hard to think of using anything else. Its only been 2 days!
My uni machine appears to be more heavily firewalled than usual... I cannot connect in or out on (amongst others) port 22, so I'm having to bounce through another machine.
My brother was having connection problems to the internet at the same time. Turns out I had left a BNC connector un-terminated when I removed my desktop from one LAN. The family computer resides on the same LAN and thus was busticated. Unfortunately he made the situation worse by resetting the gateway - so when we did work out that the BNC wasn't terminated, the firewall rules _also_ prevented access.
A little look at my christmas acquisitions part 1: Greg Egan. Off to see LOTR3 tonight. Paycheck is coming out sometime in march; there is also a new Richard Morgan book due then. A new KSR came out last monday.
Well I'm back for the time being... so I'll rant about the ridiculous state of affairs with regards to computing research. Online or Invisible? is a paper which states that online academic papers are more highly cited than ones that are only available in journals or conference proceedings. Yet, whenever I try and follow a citation from lecture notes, or papers, time and time again I find that I can't get hold of the papers in question, either via the web or academic databases.
CS is a case of `do as I say, not what I do'.
Paycheck looks like an interesting movie, but I cannot find any release information for the UK. Maybe it will not come out over here?
There are also a few additions to the site: the games section, a `mail' section under unix, and a new program on the code page.
Another 'how did I live without it' program (sortof) - icecream, a non-interactive shoutcast stream downloader. Save a stream to disk, or pipe it through to a player (which need not be able to stream itself).
This program has just been ITP'd for debian.
I've been pointed at this rather poor argument against tabs in window managers. It seems to boil down to `I don't like them so they shouldn't be implemented'. The author also seems to believe that each individual app implementing tabs to handle what are effectively multiple windows is a much better/more UNIXy solution than just implementing it once in that application that is rather well known for managing windows, yes, the window manager.
The first in a series of 'how did I live without it?' tools - wmcliphist. Wmcliphist is a small windowmaker dockapp which keeps a history of the X clipboard. You can configure it to keep a varying length of things, and can define actions associated with regular expressions against the clipboard contents. So, you can tell it to automatically start a web browser if it recognises a web URL (although you probably wouldn't want it to); or to 'file' certain things under sub-menus (that is more useful).
WMCliphist is available in Debian testing and unstable, and is far less annoying than msoffice's clipboard-history feature, which constantly pops up when I am trying to work.
Media consumed lately: Blur - Think Tank; Muse - Absolution; Simon Singh - The Code Book; Simon Singh - Fermat's Last Theorem; VA/Harlan Ellison - Dangerous Visions; Orson Scott Card - Speak for the Dead; Chuck Palahnuik - Fight Club; Greg Egan - Daispora; Rubini/Corbet - Linux Device Drivers; Spinellis - Code Reading - the Open Source Perspective; Speilberg - Catch me if you Can.
I've added recursive-categories to the links thing. I'm not too fond of the interface as it seems to take up too much vertical room. Ideally, the list of categories would be styled similarly to the list of Contents for the whole site, etc. This means I'd have to move some style information from the id's #contents and #style to some common classname for such things (navigation I've called them) and re-write all the XHTML and CSS.
Pays to do it right first time
... I wrote that and then thought "wouldn't it be a good idea to have the sub-categories extend from 'links' in the Contents, as a sub-list? All the way back to the root list, similar to how the contents expands on the St John's JCR website?" That will require significantly fewer changes...
I don't have enough stuff on this site to justify the amount of time I spend writing it. Especially code. I hope to have some SDL snippets and some AI snippets up soon, as well as my PWM patches. Functionally speaking this site is less featureful than my previous one, and that was very primitive internally in comparison.
This just in- I got a bookmarklet working for IE that lets you add bookmarks to my DB thing, which is nice. Discovered a similar thing written in PHP called 'bookmarks4u' or similar. Worth looking at (apart from the php bit). Getting tired of various aspects of ruby, perhaps time to give python a look.
Reading Greg Egan's 'Diaspora' - very good book.
I recently read an IBM developer-works article about making a Linux machine's startup procedure faster, by running startup scripts in parallel. I thought this was a very good idea, and the proposed solution sounded novel as it made use of existing software (namely, make).
I decided to see if there had been discussion about implementing such a system in Debian. Indeed there had- and it started 5 years before the IBM article was published, in 1998. The proposal seemed to have been met well by developers, so I'm wondering what fatal flaws have kept it from being implemented.
The discussion has raged on since then. 2002 suggests that Henrique de Moraes Holschuh has been working on it. Also mentioned is the proper way to go about a proposal - Proof of Concept and transition plan. Hmh's page contains details of improving the init system: http://people.debian.org/~hmh/.
A program called 'serel' was written to attempt this feat, and is detailed in a -devel thread from 2002 as well.
Oh, there's also the slashdot story, which lead to a debian-devel thread from 2000; details of a program called minit; a document by Richard Gooch entitled Linux Boot Scripts; an implementation called quickinit (rename to .tar.gz).
19:04 < michael> oh man 19:05 < michael> at the doctor's office 19:05 < michael> and they have open 802.11 19:05 < michael> so sweet
What else I'm up to: working on getting debian onto my mother's laptop; writing a guide to web design using web technologies properly; adding a games section to the site, and rambling about why I don't like PHP.
None of these are finished yet.
Just thought I'd write a quick update because that last one is annoying me.
I've been working on the links page quite a lot recently. I put together a little owner-page which lets me add links quickly and easily. I'm writing this using ruby and libdbi-ruby, and I've had some thoughts about just how useful a dbi is (more on this later.. but here's a quick snippet: not very).
Yesterday I added categories, although there are only two (well technically one): unsorted and 'IBMer's for now. Today I plan to refine the category stuff, implement sub-categories and get started on a 'last modified' property, to replace the defunct newsdot.
One final plan I have after that is to write a few browser plugins. Mozilla/NS6+ keep their bookmarks in a single .html file. I thought it would be quite cool to write a plugin which would add a menu entry for 'Remote Bookmarks' (i.e. the bookmarks file was stored on a webserver somewhere). I'd also have to write a script to churn out a compatible file from the database contents. IE would be a bit tricker, maybe a similar bit of functionality could be written using bookmarklets. Konqueror uses a single file too, so that wouldn't be a problem.
But what would be _really_ cool is if I could make the browsers transparently update the links database too, via their existing interfaces...
A programming book is like giving someone a hammer. Without showing them that you can use it to hit nails, where do they go from there?
I'm learning a lot from poking around the linux kernel internals. I think the famous Simon Howard learned most of his C from looking at carmack's doom source code. It puts K&R, Knuth or even Schildt into perspective.