W3C home > Mailing lists > Public > www-tag@w3.org > June 2008

RE: [widgets] Widgets URI scheme

From: Larry Masinter <LMM@acm.org>
Date: Sat, 14 Jun 2008 00:28:36 -0700
To: "'Marcos Caceres'" <marcosscaceres@gmail.com>
Cc: "Jon Ferraiolo" <jferrai@us.ibm.com>, "Larry Masinter" <LMM@acm.org>, <public-appformats@w3.org>, <public-appformats-request@w3.org>, <www-tag@w3.org>
Message-ID: <000001c8cdf0$42b60790$c82216b0$@org>

> Can you advise the WebApps working group (AKA WAF) on how you might like
us to proceed in regards to this matter?

Perhaps this is just a problem in the writeup that you allow both cp437 and
utf8-range?

General principles:

The purpose of standards is interoperability. 

You don't assure interoperability when there are options, unless every
reader is required to accept and properly interpret every option which a
writer is allowed to write. Otherwise, some writers will write what some
readers can't read.

In some cases, there is some communication or negotiation or ability for a
writer to discover the capabilities of the reader, and in such a case, there
might be options that aren't mandatory to implement (MTI) in readers, as
long as there is at least one option which is MTI for both readers and
writers -- a writer can otherwise only reliably use options that are
discovered among the reader's capabilities.  

In general, this doesn't usually apply to file formats which are written
outside a protocol, and where the reader isn't known when the writer is
writing a file.

In the particular case of the "widget" URI scheme proposal, you have:

    allowed-chars    = cp437 / utf8-range 

If you don't mandate that all readers must read utf8-range, but you allow
writers to create URIs with utf8-range, there will be some URIs written that
some URI interpreters can't resolve, and no easy way to tell which are which
or distinguish them.

In general, to create an interoperable specification, you either need to
restrict the writers (i.e., only use cp473) or mandate full implementations
in the writers (i.e., only use utf8).

Larry


> -----Original Message-----
> From: Marcos Caceres [mailto:marcosscaceres@gmail.com]
> Sent: Wednesday, June 11, 2008 7:08 PM
> To: Larry Masinter
> Cc: Jon Ferraiolo; Larry Masinter; public-appformats@w3.org; public-
> appformats-request@w3.org; www-tag@w3.org
> Subject: Re: [widgets] Widgets URI scheme
> 
> Hi Larry,
> 
> On Thu, Jun 12, 2008 at 11:56 AM, Larry Masinter <masinter@adobe.com>
> wrote:
> > I'm still disturbed by an interoperability problem if there are
> alternative ways of expressing the same path where not all
> implementations can read and processs all of the expressions.
> 
> I'm sure when this would happen (unless there was a broken
> implementation of Zip that encoded file names as UTF-8 without
> explicitly setting general purpose bit 11)
> 
> > The URI decoder is separate from the zip software, so its reasonable
> to use UTF8 everywhere.
> 
> Yes, this is what will be in the widget spec.
> 
> > This is very related to the issues with the file: scheme that I've
> started looking at again.
> 
> Can you advise the WebApps working group (AKA WAF) on how you might
> like us to proceed in regards to this matter?
> 
> Kind regards,
> Marcos
> --
> Marcos Caceres
> http://datadriven.com.au
> http://standardssuck.org
Received on Saturday, 14 June 2008 07:29:46 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Thursday, 26 April 2012 12:47:58 GMT