W3C home > Mailing lists > Public > public-webapps@w3.org > April to June 2009

Re: [widgets] P&C Last Call comments, interoperability

From: Marcos Caceres <marcosc@opera.com>
Date: Tue, 02 Jun 2009 12:49:47 +0200
Message-ID: <4A2503CB.6040408@opera.com>
To: Marcin Hanclik <Marcin.Hanclik@access-company.com>
CC: "public-webapps@w3.org" <public-webapps@w3.org>


On 6/2/09 12:18 PM, Marcin Hanclik wrote:
> Hi Marcos,
>
>>> No, they are different...
>  From the interoperability perspective both are "potential interoperability issues".
>
>>> You can't dictate those rules in the spec because they are based on
>>> the capability of the device.
> Of course, not everything can be captured.
> But I think the spec has already some overview of what could be a potential - and already easy to fix - interoperability issue.
> One of the initial items is e.g. length of path for the files within widget package.

No, i think HTTP tried this trick with the length of URLs. Also, look at 
what happened with Cookies having a 4Kb limit. No limits should be 
applied by the spec.

> I understand that the implementations may have arbitrary path lengths.
> But to ensure the interoperability from the very beginning, some reasonable limit could be put already.
> E.g. 1024 bytes for the maximum path length.

No, Scott Wilson's paths don't map to thing on a hard drive, they map to 
resources on the Web, where path length is irrelevant.

>>> For example, there might be devices out
>>> there that can handle a HTML canvas at size 32,000 x 32,000.
> I do not want to specify a limit in P&C for items outside of P&C spec.
> Let's specify what we can.
>
> BTW: If we already talk about future (specs to be valid in 100 years or so), there is already a limit put on HTML canvas:
> interface HTMLCanvasElement : HTMLElement {
>             attribute unsigned long width;
>             attribute unsigned long height;
> ...
> };
> but we ONLY THINK that it is limitless and unrestricted.
> Unsigned long ranges from 0 to 4294967295.
> If I would produce now a device with bigger resolution I would face an issue.

Correct, but that kinda validates my point. Given the current growth 
rate of screen resolutions and More's law, you can probably guesstimate 
when computers will be able to manipulate a canvas that is 4294967295 x 
4294967295 x 8bits/per pixel. That gives you, 1.475739525209569362 x 
10^20 (or 16 384 petabytes)... in other words, I think we are pretty 
safe with those longs probably till the sun burns out... though future 
off-world colonies may wonder why we imposed such a small limit on HTML5 
canvas:)
Received on Tuesday, 2 June 2009 10:50:26 GMT

This archive was generated by hypermail 2.3.1 : Tuesday, 26 March 2013 18:49:31 GMT