Re: Icons and document-types

["Sergey I. Panov" <sipan mit edu>]

Another "sane" message on this subject:

Jim:
> Conceptually, for me, icons are easy to do:
> 
> 1) First we identify the document type for a document.
>        (returning an "object")
> 
> 2) Then we query that object for the icon.
> 
> The object that we query will be smart enough to figure out which icon
> to use. ...

I see it as an improvement over KDE scheme, which in turn is an
improvement over MS way of doing it (I think they use magic types as a
backup for suffix based document type detection).

In KDE "Objects" with information about particular document type are
stored in "mimelnk" files. User can have his own "mimelnk" files,
which supposedly overwrite system-wide settings.

 Executables there are treated in a slightly different way. I guess,
they just trys to extract icon out of the file. A few of their
executable actually have one embeded in them, and they do not extract
icons out of non-kde programs (e.g. netscape).

 I guess the way they should be dealt with is to treat executables as
"kde-links"/"gnome-links"/"objects". E.g. if you copy executable to
the Desktop or Panel Menu the default "object" should be created there
with startup command, default icon, etc. Life can be made easier if
there is there was a data base of the default "objects" ("kde-links"
in case of KDE) for the most frequently used X applications. As soon
as such "object" ends up on the Desktop ot on the Panel Menu it can be
customized and live its own life (be different from the one in the
default "objects" data-base).

======
Yet another opinion from the one who is too lazy/buisy to code for
Humanity. Not proud but ashamed ...

  Sergey

PS: Isn't that area under tight control of the gmc team/author
(M.d.I)?  I wander what Miguel thinks about it?