[om] [IDEA] om urls

[Paul Libbrecht]

On Wednesday, January 24, 2001, at 12:41 PM, David Carlisle wrote:

> This works because telnet: is an IETF registered URI scheme. 
> The chance that you'd get om: as a registered scheme (and widely 
> understood in client browsers) seems slim. 
>  
> Mime types should also be registered, but unlike URI schemes, browsers 
> can be easily customised by end users to work with ad hoc mime types. 


They can be configured just as easily on all environments I have seen.
For example FileMaker has such a thing.

And on the Mac, this can even happen in the complete background (Internet Config Settings and register URLHandler). In Windows, this probably.... the Windows Registery (!).


> I'm not at all sure this is desirable anyway. You are encoding the XML 
> content of the expression inside the URI, this works up to a point, but 
> probably most URI handlers are expecting a relativly short URI pointing at 
> relativly long documents. You could just as easily encode an HTML 
> document inside a URI, but it's more common to put the HTML on the web 
> and point at it with a URI.... If you just use linking using a standard 
> scheme such as http, your browser will fire off a suitable application 
> if the entity returned has a mime type that is associated with that 
> application. 

This point is definitely valid. I have not searched yet about these limits, the OM binary encoding might be an escape (as the URL is almost anyways unreadable).
Note however that huge URLs are getting more and more popular as the web evolves towards rich applications.


>  
> The proposal for XML mime types 
> ftp://ftp.isi.edu/in-notes/rfc3023.txt 
> would have the mime type as application/openmath+xml. If you 
> serve .om files as application/openmath+xml then a browser 
> coming accross an http (or ftp or any other retrieval scheme) 
> link will use whatever helper application is associated with 
> that mime type in your browser's configuration or .mailcap file. 
>  

This works (and would of course be done by more integrated schemes) and OMcentral would also register itself for such, probably.

Now, the thing would not work all the time as it would need each open-math-object to be individually reachable (by a request like protocol). The protocol would then remain anything else (like http, ftp, file or whatever such), this would also involve OMcentral to handle such protocol (for those quoted, this is kind of OK but then... imagination is pretty far reaching).

It would be much less portable then. 
For example, it would not be suitable for local file-exchange, email-exchange or so and that point is major to me. This has to be document-based.

These two mechanisms (the pull and get a given MIME-type) and the in URL content.

Paul
--
om at openmath.org  -  general discussion on OpenMath
Post public announcements to om-announce at openmath.org
Automatic list maintenance software at majordomo at openmath.org
Mail om-owner at openmath.org for assistance with any problems