Re: [widget] Remarks on http://www.w3.org/TR/2007/WD-widgets-20071013/

Hi Mohamed,
Thanks for taking the time to review the spec.

On 10/23/07, Innovimax SARL <innovimax@gmail.com> wrote:
>
> http://www.w3.org/TR/2007/WD-widgets-20071013/
>
> In 2.1. Widget User Agents
> Why is XML 1.0 between square bracket ?

It was to be a citation. I'll fix up the referencing as the document matures.

> May be I come late in the battle, but not sure to understand why ZIP
> is mandated instead of zlib/gzip which is already implemented for
> handling PNG and HTTP 1.1 (which are both mandatory if I'm not wrong)?

I don't think there will be a battle. Gzip is incompatible with Window
XP/Vista Compressed Folders so that rules it out straight away.
Secondly, gzip can only compress one file (not a collection of files),
so you would also need something like a tar ball [1] (hence,
*.tar.gz). Tar is also not natively available on Windows systems; and
even though tar and gzip are on MacOS, it would require authors to use
command line tools to create a widget package (which goes against our
"ease of use" design goal [2]). Lastly, no widget engine is using
gzip, and hence it would be going against the grain to recommend it.

As for PNG and HTTP, they both have their origin in the unix world
where gzip is widely available. Also, I read [3] that it was some of
the zlib guys that worked on PNG, hence it's use of zlib/gzip.

>
> In 3.1. Attribute Types
> Please add wether
> 00012 is a valid non negative integer, and wether
> 0000.0000 is a valid non-negative float
> One clear thing in your current definition is that ".456" and "1." are
> not valid non-negative float

I will clarify this in the spec.

> Please add a newline between "valid-version-tag :=
> version-identifier*('.'version-identifier)" and "version-identifier :=
> string" . Same point for "version attribute" paragraph

Fixed.

> Not sure to understand what are the allowed char in "string" : are
> FULL STOP allowed? SPACE ? COMMA ?

At this point, all characters are allowed....

> If it is the case, why not just say
> "valid-version-tag := string" ?

Good point, but the "." may become important as we try to resolve the
automatic updates/versioning issue. If it turns out that "." has no
significance, then valid-version-tag will simple be a string, as you
suggested.

> Are "whitespace" used in "Keyword attribute" definition the same as
> "space characters" defined earlier ?

Yes, fixed.

> Not sure to understand why "The widget Element" definition is in "3.1
> Attribute Types" ?
>
> I fear that your are going to far into detail for the XML structure.
> Even if it could be a heavy burden, please think of using XML Schema
> Part 2 (Datatypes)

I'm aware that the is a problem there as those definitions don't link
to the actual configuration file elements yet. Section "3.1 Attribute
Types" will probably vanish in a future draft. Regardless, we don't
want any normative (or otherwise) dependencies on XML Schema; that
would only complicate matters (it's bad enough that we included a
namespace).

> Other than that, it is not clear (since you introduce a special
> definition of whitespace) is those whitespace are allowed in any
> attribute value ? If so, please precise it, and the give the process
> (trim) to remove them (or transform them to space ?)
>
> Ok I see it now, it is in 3.8 : please announce it sooner

Will do.

> Reference
> * RFC2822 is referenced but without content
> * Please update XML to the fourth edition
>
> It miss a least reference for GIF, HTTP 1.1, XMLHTTPRequest, and  ECMAScript
>

Thanks, I'll add those in. However, don't expect the bibliography to
be finished until the document matures.

Thanks again for your feedback!

Kind regards,
-- 
Marcos Caceres
http://datadriven.com.au
[1] http://en.wikipedia.org/wiki/Tar_(file_format)
[2] http://www.w3.org/TR/widgets-reqs/
[3] http://www.info-zip.org/Info-ZIP2.html

Received on Wednesday, 24 October 2007 03:34:49 UTC