URI DEV: Difference between revisions

Revision as of 13:16, 1 February 2008

General

The current URI system is a mess made up of different sets of parameters for the same purpose depending on who and when wrote somehting. the idea is to first to remove unneccessary stuff from it, second standarize parameter and third make ir richer in feature

Guidelines

URI's:

don't change
are human guessable
are logical (no need to mirror a filesystem)
help visualize the site structure
are short
use lowercase
don't use unexpected punctuation
lack query parameters
allow public linking
are stateless
dont expose technology

(blatantly stolen from: http://www.eakes.org/archives/2005/01/flickr_uris.html )

what not to do:

use plurar for parameter when you shouldn't (i.e. animes/5 instead of anime/5)
use "=1/=on". if the parameter is in the url it IS on. if not it's not which leads to
don't add unneccessary parameter

Vision

1(!) standarized Parser through which the data passes first
unnessary scary things like /perl-bin/animedb.pl
structured URI
standarized parameter
let http://anidb.net/gonzo or http://anidb.net/chobits use the Full Text Search and return useful stuff instead of failing ([1]])
use dictionaries where possible and avoid unneccessary &
use POST insteed of GET when possible. i.e. for adds, deletes, edits, votes. basically anythign that actually modifies data. for anyhting else (read only) gets should be used
...

shortterm simple Changes

get rid of "/perl-bin/animedb.pl"

yes we use perl. no we don't need to throw our "perl penis into the face of everoyne else" (as rar calls it)

...

Design idea 1

http://anidb.net/<page>(/<id|name>/<action>/)/<paratemer>

<id|name> and <action> are only needed for specific entries hence why those pages should not use plural for the pagename i.e. anime instead of animes the plural version should be reserved for lists. so instead of animelist we just use animes. that simpler and more logic. aside shorter.

actions could be:

/show/ (could be used by default, but we should really just leave that out. unneccessary clutter. accepting it anyway would be fine though. accepting is fine. forcing is bad.)
/edit/
/history/
/vote/
/delete/

...

parameter could be something like

/h&mylist&noalias&wishlist&orderby.name.desc

Examples

this:

http://anidb.net/perl-bin/animedb.pl?h=1&mylist=1&noalias=1&orderby=name&orderdir=0&show=animelist&wishlist=1

could look like this:

http://anidb.net/animes/porn&inmylist&noalias&wishlist&orderby.name.asc

this:

http://anidb.net/perl-bin/animedb.pl?show=anime&aid=1

like this:

http://anidb.net/anime/1

and this:

http://anidb.net/perl-bin/animedb.pl?show=addanime&edit=1

becomes:

http://anidb.net/anime/1/edit

@@ Line 30: / Line 30: @@
 * let http://anidb.net/gonzo or http://anidb.net/chobits use the Full Text Search and return useful stuff instead of failing ([http://tracker.anidb.net/view.php?id=798|Tracker]])
 * use dictionaries where possible and avoid unneccessary &amp;
+* use POST insteed of GET when possible. i.e. for adds, deletes, edits, votes. basically anythign that actually modifies data. for anyhting else (read only) gets should be used
 * ...