Re: another proposal for MIME: with actions, implementators, and scores

[Lee Braiden <jel ntlworld com>]

Hi Jody,

Thanks for the input :)


On Thursday 11 Dec 2003 3:35 am, Jody Goldberg wrote:
> On Mon, Dec 08, 2003 at 10:04:54PM +0000, Lee Braiden wrote:
> A nice idea, but I suspect this would get problematic.  Assigning
> the scores would be fairly contensious.

Agreed, but that's a policy/discipline issue within the GNOME(/other desktop) 
projects.  More contentious issues than this have been handled with the HIG, 
etc :D

On the other hand, it IS a fairly well-defined rating system -- things that 
aren't implemented simply aren't implemented.  Things which implement 50% of 
features implement 50%, and so on.  That 50% implementation of features in 
the application doesn't change, even if the application's actual format 
importer understands the format completely -- 50% is still 50%.

> An example near and dear to my heart would be Gnumeric vs OpenCalc
> vs MS Excel under wine.  Each choice has its relative merits.
> Choosing a ranking would depend on what's important to a user.
>     Gnumeric	- fast, solid spreadsheet and analytics
>     OO		- native office suite
>     Excel	- The best compatibility at the expense of nativeness

I covered automatic selection of applications for your current desktop, and 
modifiers to allow for a user's preferred applications.  The 
'IDEAL_IMPLEMENTATION' was intended to handle the native application case.  
Obviously I didn't explain this clearly, but I rushed, like I said ;)

> How about things like filter quality.  Gnumeric has a solid core,
> but our lotus 123 filter could use more bells and whistles.

Covered this above to some extent.

> Do we list the scores on a per mime type basis ?

No -- on a per action, per mimetype basis.  That is, one score for how well an 
app Opens/Views HTML, another for Editing, etc.

I agree, this is a complex thing to propose, and it requires a lot of work 
from app maintainers, codec/plugin developers, etc.  But honestly... think 
about what we've got now... we have an static system that knows about 
'applications' and 'mimetypes'.  Mimetypes are currently not working anyway, 
to the best of my understanding.  Add-ons like new fileformat plugins aren't 
catered for.

As for the actual complexity of all these applications and the operations 
(open,edit,etc) that can be done on a file, and how well.... sure, the 
desktop doesn't have to worry about that right now -- the USER does!

This is a classic case of something that needs reworked, imho.  I'm not saying 
my solution is ideal (in fact, I mostly just pulled it out of my butt when 
the topic came up ;), but I REALLY want people to start thinking about 
broader solutions to the whole file/application interaction thing.

> Or even more subtle MS
> Excel's binary xls format vs their various XML formats.  Do they
> even have different mime types ?

Yes, that's an interesting point.  I've no idea on that one, personally.  All 
I can add about it is that a global preference could 'prefer' open formats IF 
enough information is provided by applications/desktop files/a new mime 
system.  Otherwise, again, the desktop is 'dumb', and just leaves it all for 
the user to cope with.

> My suspicion is that there would be lots of similar situations, and
> that the user wouldn't benefit much from the added complexity.

You could be right.  But let's not throw out ideas because they *might* have 
problems ;)

-- 
Lee.