|
start |
find |
index |
login or register |
edit
| |
2007-04-21
Peter Norvig's How to Write a Spelling Corrector recently made the rounds. In brief it's an article around 20 21 lines of Python demonstrating how to do statistical spelling correction. The article is a nice read so go ahead and read it. After having done that, consider the last section: the errata. In those 20 lines of code, 2 errors were found after the original version was published. One is basically an "internationalization"-related thing and the other one is an error in an essential part of the algorithm's implementation. A third problem in the "documentation" (Norvig's article) is also mentioned. Now that strikes me as particularly interesting: 20 lines of code, 2 errors. Don't get me wrong, I understand the intentions of the article. But this error ratio made me drifting off with some nice thoughts about code correctness, so ... maybe you want to drift off as well :)
Etch upgrade status: 1 machine done, n – 1 to go. The first machine was upgraded from unofficial amd64 sarge, the process was extremely smooth: not a single problem encountered.
Armin Ronacher: Vim as Development Environment - "I do all of my development (well, except of some Java stuff [..]) using VIM." [via deli/gavin]
Things I also use daily and which you should therefore (obviously :) take away (in no particular order):
- if you do markup, you must have closetag (CTRL-_)
- CTRL-D (the wildmenu) for command-ine completion (of e.g. filenames)
- CTRL-N for quick'n'dirty insert mode completion
- incremental search (
/ , :set incsearch , :nohl ) - keys:
. (repeat), * (search fwd), % (jump to match), < and > (shift text) - maybe you like minibufexpl - a buffer "explorer"
|
echo earlZstrainYat|tr ZY @.
|
earl.strain.at
• esa3
• online for 8653 days
• c'est un vanilla site
|