This simplifies a while loop in the js_compress() code. The
functionality of the new code is completely identical to the old code
but it uses less comparisons and according to FS#2593 it is thus a lot
faster.

See
http://stackoverflow.com/questions/805107/how-to-create-multiline-strings
for info on them.

Instead of three stylesheets for 'all', 'screen' and 'print' modes, they are all loaded into a single stylesheet by wrapping all screen styles in a "@media screen {}" and all print styles in a "@media print {}". The 'all' mode is not wrapped in anything.
Potential issues with existing CSS: If any of your screen or print CSS files already contain any "@media" syntax, the CSS will probably break. In that case please add any CSS with "@media" in it to the 'all' mode instead!

Also, the 'rtl' mode is hereby deprecated. Please just prefix any RTL styles within your normal CSS files with "[dir=rtl]". This also fixes that RTL styles cannot be added for 'all' or 'print' modes.

This removes the cachekey parameter again and instead follows
@michitux's suggestion to trigger an event for the cache usage

The score was randomly transfered as string or as integer.
This way it will always be transfered as an integer.

The score was randomly transfered as string or as integer.
This way it will always be transfered as an integer.

Added the new error codes and categories:

--- 212	Not allowed to delete media

== 230	Media edit error
--- 231	Filename not given
--- 232 File is still referenced
--- 233	Could not delete file

Since there are currently a lot of error coded returning 1, that mean
completly different thing, i guess it would be very nice to change this.
A client should not be forced to parse the error message, the error code
should be enough to explain the error.

This change suggests some error codes, that have a hierarchical
structure. In the following list the categories begin with = and the
error codes actually used with -.

= 100 	Page errors
== 110	Page access errors
--- 111	User is not allowed to read the requested page
--- 112	User is not allowed to edit the page
== 120	Page existance errors
--- 121	The requested page does not exist
== 130 	Page edit errors
--- 131	Empty page id
--- 132	Empty page content
--- 133	Page is locked
--- 134	Positive wordblock check

= 200 	Media errors
== 210	Media access errors
--- 211	User is not allowed to read media
---	215	User is not allowed to list media
== 220	Media existance errors
--- 221	The requested media does not exist

= 300	Search errors
== 310	Argument errors
--- 311	The provided value is not a valid timestamp
== 320	Search result errors
--- 321	No chances in specified timeframe

For some reason trailing/leading underscores were allowed when uploading
files. But the rest of the code (eg. listing or downloading files) never
supported this. This patch removes this special case for uploading files
to streamline ID cleaning of pages and media files.

Without creating an IXR_Base64 object, the file will be encoded as base64, but send as string. The client XML-RPC parser cannot detect that it is meant to be a base64 encoded file.