Weblog


Weblog

how to write unmaintaineable code

11.10.2002, 10:18 · Martin · Permalink

field-excursion day 11. Declare each variable five times in each bit of code, and nake sure the code won't work (in a non-obvious way) if one of the declarations is missing. Bonus-points: write wrapper functions to do the declarations that take as arguments the (slightly altered) variable names to be declared. Find some creative way to build same variable names on the fly from yet another list of variable name components, whereby cunningly altering the sequence of named parts in the argumnent lists. Can you tell what role I'm playing in this? 100 point to anyone who said 'maintainer' how to maintain unwriteable code
Curently I'm writing obfuscated Perl code to analyse unmaintainable PHP code. Go figure. In his unchecked youthfull selfoverestimation, Mr. Fink thinks I'm referring to his code, which I am not. Your code is seldom bloated and can be maintained rather well (by me), as we wrote most of that together anyway. Believe me, these guys can teach us all yet.

The Future is Wild

11.10.2002, 09:18 · Martin · Permalink

How will our earth evolve, the clima change, the geograpy? Say in 200 Mio. Jears from now?

Automatic Document Classification

10.10.2002, 09:28 · Martin · Permalink

Automatic Document Classification With Perl based on 'naive bayesian'... this seems to be exactly what I was searching for. AI::Categorize::NaiveBayes allows the user to feed it the text of several documents (the training set), which it will parse and add to the word frequency database
<snip />
Once a sufficient number of training documents have been fed to the database and the needed probabilities have been calculated, we can start asking AI::Categorize::NaiveBayes to categorize new documents that it hasn't seen before. It returns to us an ordered list of the most probable categories for that document.
TeledyN: Comment on Graham's Plan for Spam
The specific filtering of spam also reminded me of the 1994 ACM project to produce a collaborative filter to rid the dying USENET from spam attacks; http://www.si.umich.edu/~presnick/papers/cscw94/GroupLens.htm chronicals the GroupLens project, and sure enough, there's the same Bayesian method at the root of it. The ifile Web Site
ifile is a general mail filtering system that works with a mail client to intelligently filter mail according to the way the user tends to organize mail. ifile uses the machine learning algorithm Naive Bayes to classify e-mail documents. Freely Available Filtering Systems, Information Filtering Resources Personal WebWatcher, Project Page
Personal WebWatcher is a "personal" agent that accompanies you from page to page as you browse the web, highlighting hyperlinks that it believes will be of interest. Its strategy for giving advice is learned from feedback from earlier tours.

research (reading list)

09.10.2002, 21:43 · Martin · Permalink

The Adaptive Resonance Theory (ART) clearinghouse A Plan for Spam (Bayesian in Perl im ARC, which is a kind of Lisp)
Spam Detection (expanding upon Grahams article) Various bits of software: SpamOracle
Bayesian filter in O'Caml

Publications of David Mertz

09.10.2002, 08:41 · Martin · Permalink

A version of this page with essentially the same markup was written back in the days when folks like me were complaining about the new-fangled features of HTML 2.0, and about the associated erosion of markup for structural/semantic meaning into markup for visual appearance. I think the page holds up pretty well after eight years (even visually). In fact, I will make the very small changes necessary to make this compliant XHTML, and give it another decade of life. Text processing in Python
(Book in progress and freely online)
has a chapter on Parsers and Staemachines which looks supiciously like what I need for my PHP parsing stuff

come read my email

09.10.2002, 08:24 · Martin · Permalink

Ray Ozzie muses about making all email communication public. Well, not all, but all official. It's an intersting argument he has there. Why not communicate out in the open? And one thought that occured to me when reading that... People do write more and more freely if they have a certain person to adress. Writing down thoughts is more difficult than writing down a real request or response.

developerWorks: Six approaches to eliminating unwanted e-mail

08.10.2002, 22:33 · Martin · Permalink

this article has a real good feel to it. Somehow I rather like the way the way the Bayesian techniques cut better than the other aproaches on several fronts...

CongaRatz!

08.10.2002, 20:53 · Martin · Permalink

to my friend Andi who is finally for hire if you employ less than 50 other people already. Give that man some work, keep him busy or he'll be out in the sun of sunny Spain too long which would not be good for his teint

hellocode : CODE

08.10.2002, 10:59 · Martin · Permalink

UI prototypes in FLASH... talk about ZUI, huh?
[via gavin]

When I see digital art

08.10.2002, 09:09 · Martin · Permalink

like this or this I want to pack up and go home.

how mark did his 'good and easy'

08.10.2002, 08:17 · Martin · Permalink

There's a lot usefull stuff to learn here. YMMV, but do try some of that stuff

play abalone online

08.10.2002, 07:57 · Martin · Permalink

looking for a real quality time-waster? Try playing abalone online, free and against live opponents, Really nice made Macromedia Director game. Thrill from the official abalone maker

what a cutie!

07.10.2002, 08:53 · Martin · Permalink

my has Ian grown... I still thought of him in terms of the first pictures I ever saw... which was almost a year ago...

it took a lot of work

06.10.2002, 23:22 · Martin · Permalink

for this to look no different. It's under the hood that things are a changing. And a lot at that. I am again one step further away from the Frontier-website-framewirk rendering process, and on my road to full Traumtank-ness. Re-creating functionality to dynamically build folder indices and such...

das klingt aber nicht gut

04.10.2002, 12:04 · Martin · Permalink

auch wenn ich nur zu gut weiß wie sich das anfühlt.

Google degregation...

03.10.2002, 08:28 · Martin · Permalink

Well, as many people noted, Google has changed something in their ranking. But while most relevant websites notice a decline in search-hits, my humble site saw a flurry of new hits for "cd cover" and "zeichnungen" and such (they have always been high). Now I'm not complaining about this publicity, but I also don't thinks thats a plus in relevant searches...

my, now that was refreshing

03.10.2002, 00:26 · Martin · Permalink

It seems doing some artwork in PS for a change did wonders to my tierd and unethuseastic self. So now I'm all powered up and energetic, only to find it's past midnight (again). I could keep on for some time now... Well, not so bad, staying home the next few days.

AngelCloud

02.10.2002, 23:48 · Martin · Permalink

getting completly carried away in PhotoShop has sometimes nice effects, if you consider this cover took about three hours max from start to end. (okay, I already had that angel lying around somewhere) Only problem is, garret's eMail with the original high-res cloud-photo went un-noticed in my inbox. So what you see here is made with the small one he posted to his website earlier...

'twas time I RTFM

02.10.2002, 15:42 · Martin · Permalink

... I just don't get ht://Dig indexing... how do I index incrementaly? I just have a few new pages per day, the others hardly ever change... one of these days I'll need too read that manual ;) So for now, you just as well ignore the search-box on youre right...

those clouds

02.10.2002, 15:32 · Martin · Permalink

garret 's beeen posting lately.. I'd love to have hi-res copies of them ;)


alles Bild, Text und Tonmaterial ist © Martin Spernau, Verwendung und Reproduktion erfordert die Zustimmung des Authors

Martin Spernau
© 1994-2026


amazon.de Wunschliste

Facebook me!
Google+

powered by Traumtank