From stevelist at silverorange.com  Fri May  1 06:44:10 2009
From: stevelist at silverorange.com (Steven Garrity)
Date: Fri, 01 May 2009 10:44:10 -0300
Subject: [html5] <video> tag fallback
In-Reply-To: <06DD1FDB-82CC-49B7-8E0D-60B62CA459BD@apple.com>
References: <49E47F2F.2030709@silverorange.com>
	<0FD246DD-3C30-40D8-89F8-7FFF75C3A257@iki.fi>
	<49E4CA2F.7000800@ucsc.edu>
	<DD0C1FCA-09A0-4FCB-AE27-68295F0CDDD9@apple.com>
	<Pine.LNX.4.62.0904300616260.10370@hixie.dreamhostps.com>
	<06DD1FDB-82CC-49B7-8E0D-60B62CA459BD@apple.com>
Message-ID: <49FAFCAA.6040405@silverorange.com>

Eric Carlson wrote:
>   WebKit nightly builds do fire an 'error' event if no suitable src is 
> found.

That's encouraging, but it seems that existing releases of Safari 3.x 
already attempt to support the <video> tag without such an error event. 
So, implementing for real users is already difficult.

If we want a solution that isn't broken for existing Safari users, we'll 
probably have to resort to a browser-detection hack.

Also, throwing a JS exception isn't helpful for users without JavaScript 
enabled.

Not trying to sound like a complainer - just want to make sure you're 
aware of what actual implementers are running into.

Thanks,
Steven Garrity

From caoyp at cn.fujitsu.com  Tue May 19 18:27:24 2009
From: caoyp at cn.fujitsu.com (caoyipeng)
Date: Wed, 20 May 2009 09:27:24 +0800
Subject: [html5] a question about canvas element's shadow effect
Message-ID: <2E9A38A0754E4E22B85E659E028883ED@FNSTcyp>

canvas element's shadow effect use Gaussian blur to fulfill.

In Gaussian blur, when do convolution operation, Gaussian template's size(or
Gaussian kernel width and height) affect the blur effect

,is there default value about Gaussian template's size(such as the size is
(6?+1)*(6?+1),? is the standard deviation).

 

Regards
C.Y.P.

 

Cao Yipeng
Development Dept.I
Nanjing Fujitsu Nanda Software Tech. Co., Ltd.(FNST)
8/F., Civil Defense Building, No.189 Guangzhou Road ,
Nanjing , 210029, China 
TEL: +86+25-86630566-817
COINS: 79955-817
FAX: +86+25-83317685
MAIL: caoyp at cn.fujitsu.com

 

 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.whatwg.org/pipermail/help-whatwg.org/attachments/20090520/5479f90b/attachment.htm>

From sean at miscoranda.com  Mon May 25 09:14:39 2009
From: sean at miscoranda.com (Sean B. Palmer)
Date: Mon, 25 May 2009 17:14:39 +0100
Subject: [html5] Microdata and Meta
Message-ID: <b6bb4d890905250914x41f3c2a1q12a5d58cd89a22ad@mail.gmail.com>

I know the Microdata stuff isn't stable yet, but I was just wondering
if the following is a valid and sensible thing to do:

<!DOCTYPE html>
<head item>
<title>Microdata Example</title>
<link itemprop="about" href="">
<meta itemprop="example" content="value">

The idea is that meta/@itemprop is somewhat like a user space version
of meta/@name, but to enable the former to work exactly like the
latter you have to make it clear that the item being referred to is
the current document. Hence the <link> element in the recipe above.

If that's valid, I wonder if it'd be worth making the following equivalent:

<!DOCTYPE html>
<title>Microdata Example</title>
<meta itemprop="example" content="value">

Mainly because when you use <meta> in head, you already expect it to
be giving metadata about the present document. The item Microdata
construct is basically extending the semantics of <meta>, and when you
use it in its old capacity I wonder if people will get confused. Maybe
not, and they should read the specifications and tutorials properly
anyway; but this is the kind of common sense way I'd expect it to
work.

Also, I must admit that I am prone to making up my own <meta> names
and values. Most of this stuff is half way between being private and
public data, and definitely aligns to the traditional semantics of
<meta> enough to make an @data-* attribute seem like the wrong thing
to do. But I don't know how universal these use cases are.

-- 
Sean B. Palmer, http://inamidst.com/sbp/

From sean at miscoranda.com  Mon May 25 09:24:15 2009
From: sean at miscoranda.com (Sean B. Palmer)
Date: Mon, 25 May 2009 17:24:15 +0100
Subject: [html5] Absolute Pixels in HTML 5
Message-ID: <b6bb4d890905250924n27b1d457m4cda9418614b837b@mail.gmail.com>

HTML 5 currently says that img/@width and so forth use CSS 2.1 pixel
values. Pixel values in CSS 2.1 are of course a relative rather than
an absolute unit. In other words, CSS px is a non-linear alias for
radians; 675.522px, for example, is a quarter of a radian.

But non-vector image formats like JPG and PNG have discrete pixels,
and screens have pixel displays with certain ppi values. So has
anybody looked at whether browsers actually implement px as a relative
unit? And how do they render images? I'm presuming that they don't
resample to fit the relative measurement.

>From the point of view of an author, I'm mainly concerned with, for
example, setting widths for figures. If I'm not giving any height or
width attributes for an <img> and my browser is rendering the image
absolutely, I don't then want to set a relative width for the figure.

-- 
Sean B. Palmer, http://inamidst.com/sbp/

From foolistbar at googlemail.com  Mon May 25 10:45:03 2009
From: foolistbar at googlemail.com (Geoffrey Sneddon)
Date: Mon, 25 May 2009 18:45:03 +0100
Subject: [html5] Absolute Pixels in HTML 5
In-Reply-To: <b6bb4d890905250924n27b1d457m4cda9418614b837b@mail.gmail.com>
References: <b6bb4d890905250924n27b1d457m4cda9418614b837b@mail.gmail.com>
Message-ID: <F6A515C3-1284-46A6-A592-CBE3615E665D@googlemail.com>


On 25 May 2009, at 17:24, Sean B. Palmer wrote:

> HTML 5 currently says that img/@width and so forth use CSS 2.1 pixel
> values. Pixel values in CSS 2.1 are of course a relative rather than
> an absolute unit. In other words, CSS px is a non-linear alias for
> radians; 675.522px, for example, is a quarter of a radian.
>
> But non-vector image formats like JPG and PNG have discrete pixels,
> and screens have pixel displays with certain ppi values. So has
> anybody looked at whether browsers actually implement px as a relative
> unit? And how do they render images? I'm presuming that they don't
> resample to fit the relative measurement.
>
>>
> From the point of view of an author, I'm mainly concerned with, for
> example, setting widths for figures. If I'm not giving any height or
> width attributes for an <img> and my browser is rendering the image
> absolutely, I don't then want to set a relative width for the figure.

What HTML 5 defines is what browsers currently do, but it's also true  
that all desktop browsers set 1 CSS pixel = 1 physical pixel, so the  
issue doesn't really come up, though the definition matches what  
happens in cases where 1 CSS pixel does not equal 1 physical pixel.


--
Geoffrey Sneddon
<http://gsnedders.com/>


From sean at miscoranda.com  Tue May 26 10:30:28 2009
From: sean at miscoranda.com (Sean B. Palmer)
Date: Tue, 26 May 2009 18:30:28 +0100
Subject: [html5] Absolute Pixels in HTML 5
In-Reply-To: <F6A515C3-1284-46A6-A592-CBE3615E665D@googlemail.com>
References: <b6bb4d890905250924n27b1d457m4cda9418614b837b@mail.gmail.com>
	<F6A515C3-1284-46A6-A592-CBE3615E665D@googlemail.com>
Message-ID: <b6bb4d890905261030j737cf228t113fdd9b78e8489@mail.gmail.com>

On Mon, May 25, 2009 at 6:45 PM, Geoffrey Sneddon wrote:

> What HTML 5 defines is what browsers currently do, but it's also true that
> all desktop browsers set 1 CSS pixel = 1 physical pixel, so the issue
> doesn't really come up, though the definition matches what happens in cases
> where 1 CSS pixel does not equal 1 physical pixel.

?

-- 
Sean B. Palmer, http://inamidst.com/sbp/

From stevelist at silverorange.com  Fri May  1 06:44:10 2009
From: stevelist at silverorange.com (Steven Garrity)
Date: Fri, 01 May 2009 10:44:10 -0300
Subject: [html5] <video> tag fallback
In-Reply-To: <06DD1FDB-82CC-49B7-8E0D-60B62CA459BD@apple.com>
References: <49E47F2F.2030709@silverorange.com>
	<0FD246DD-3C30-40D8-89F8-7FFF75C3A257@iki.fi>
	<49E4CA2F.7000800@ucsc.edu>
	<DD0C1FCA-09A0-4FCB-AE27-68295F0CDDD9@apple.com>
	<Pine.LNX.4.62.0904300616260.10370@hixie.dreamhostps.com>
	<06DD1FDB-82CC-49B7-8E0D-60B62CA459BD@apple.com>
Message-ID: <49FAFCAA.6040405@silverorange.com>

Eric Carlson wrote:
>   WebKit nightly builds do fire an 'error' event if no suitable src is 
> found.

That's encouraging, but it seems that existing releases of Safari 3.x 
already attempt to support the <video> tag without such an error event. 
So, implementing for real users is already difficult.

If we want a solution that isn't broken for existing Safari users, we'll 
probably have to resort to a browser-detection hack.

Also, throwing a JS exception isn't helpful for users without JavaScript 
enabled.

Not trying to sound like a complainer - just want to make sure you're 
aware of what actual implementers are running into.

Thanks,
Steven Garrity


From caoyp at cn.fujitsu.com  Tue May 19 18:27:24 2009
From: caoyp at cn.fujitsu.com (caoyipeng)
Date: Wed, 20 May 2009 09:27:24 +0800
Subject: [html5] a question about canvas element's shadow effect
Message-ID: <2E9A38A0754E4E22B85E659E028883ED@FNSTcyp>

canvas element's shadow effect use Gaussian blur to fulfill.

In Gaussian blur, when do convolution operation, Gaussian template's size(or
Gaussian kernel width and height) affect the blur effect

,is there default value about Gaussian template's size(such as the size is
(6?+1)*(6?+1),? is the standard deviation).

 

Regards
C.Y.P.

 

Cao Yipeng
Development Dept.I
Nanjing Fujitsu Nanda Software Tech. Co., Ltd.(FNST)
8/F., Civil Defense Building, No.189 Guangzhou Road ,
Nanjing , 210029, China 
TEL: +86+25-86630566-817
COINS: 79955-817
FAX: +86+25-83317685
MAIL: caoyp at cn.fujitsu.com

 

 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.whatwg.org/pipermail/help-whatwg.org/attachments/20090520/5479f90b/attachment-0001.htm>

From sean at miscoranda.com  Mon May 25 09:14:39 2009
From: sean at miscoranda.com (Sean B. Palmer)
Date: Mon, 25 May 2009 17:14:39 +0100
Subject: [html5] Microdata and Meta
Message-ID: <b6bb4d890905250914x41f3c2a1q12a5d58cd89a22ad@mail.gmail.com>

I know the Microdata stuff isn't stable yet, but I was just wondering
if the following is a valid and sensible thing to do:

<!DOCTYPE html>
<head item>
<title>Microdata Example</title>
<link itemprop="about" href="">
<meta itemprop="example" content="value">

The idea is that meta/@itemprop is somewhat like a user space version
of meta/@name, but to enable the former to work exactly like the
latter you have to make it clear that the item being referred to is
the current document. Hence the <link> element in the recipe above.

If that's valid, I wonder if it'd be worth making the following equivalent:

<!DOCTYPE html>
<title>Microdata Example</title>
<meta itemprop="example" content="value">

Mainly because when you use <meta> in head, you already expect it to
be giving metadata about the present document. The item Microdata
construct is basically extending the semantics of <meta>, and when you
use it in its old capacity I wonder if people will get confused. Maybe
not, and they should read the specifications and tutorials properly
anyway; but this is the kind of common sense way I'd expect it to
work.

Also, I must admit that I am prone to making up my own <meta> names
and values. Most of this stuff is half way between being private and
public data, and definitely aligns to the traditional semantics of
<meta> enough to make an @data-* attribute seem like the wrong thing
to do. But I don't know how universal these use cases are.

-- 
Sean B. Palmer, http://inamidst.com/sbp/


From sean at miscoranda.com  Mon May 25 09:24:15 2009
From: sean at miscoranda.com (Sean B. Palmer)
Date: Mon, 25 May 2009 17:24:15 +0100
Subject: [html5] Absolute Pixels in HTML 5
Message-ID: <b6bb4d890905250924n27b1d457m4cda9418614b837b@mail.gmail.com>

HTML 5 currently says that img/@width and so forth use CSS 2.1 pixel
values. Pixel values in CSS 2.1 are of course a relative rather than
an absolute unit. In other words, CSS px is a non-linear alias for
radians; 675.522px, for example, is a quarter of a radian.

But non-vector image formats like JPG and PNG have discrete pixels,
and screens have pixel displays with certain ppi values. So has
anybody looked at whether browsers actually implement px as a relative
unit? And how do they render images? I'm presuming that they don't
resample to fit the relative measurement.

>From the point of view of an author, I'm mainly concerned with, for
example, setting widths for figures. If I'm not giving any height or
width attributes for an <img> and my browser is rendering the image
absolutely, I don't then want to set a relative width for the figure.

-- 
Sean B. Palmer, http://inamidst.com/sbp/


From foolistbar at googlemail.com  Mon May 25 10:45:03 2009
From: foolistbar at googlemail.com (Geoffrey Sneddon)
Date: Mon, 25 May 2009 18:45:03 +0100
Subject: [html5] Absolute Pixels in HTML 5
In-Reply-To: <b6bb4d890905250924n27b1d457m4cda9418614b837b@mail.gmail.com>
References: <b6bb4d890905250924n27b1d457m4cda9418614b837b@mail.gmail.com>
Message-ID: <F6A515C3-1284-46A6-A592-CBE3615E665D@googlemail.com>


On 25 May 2009, at 17:24, Sean B. Palmer wrote:

> HTML 5 currently says that img/@width and so forth use CSS 2.1 pixel
> values. Pixel values in CSS 2.1 are of course a relative rather than
> an absolute unit. In other words, CSS px is a non-linear alias for
> radians; 675.522px, for example, is a quarter of a radian.
>
> But non-vector image formats like JPG and PNG have discrete pixels,
> and screens have pixel displays with certain ppi values. So has
> anybody looked at whether browsers actually implement px as a relative
> unit? And how do they render images? I'm presuming that they don't
> resample to fit the relative measurement.
>
>>
> From the point of view of an author, I'm mainly concerned with, for
> example, setting widths for figures. If I'm not giving any height or
> width attributes for an <img> and my browser is rendering the image
> absolutely, I don't then want to set a relative width for the figure.

What HTML 5 defines is what browsers currently do, but it's also true  
that all desktop browsers set 1 CSS pixel = 1 physical pixel, so the  
issue doesn't really come up, though the definition matches what  
happens in cases where 1 CSS pixel does not equal 1 physical pixel.


--
Geoffrey Sneddon
<http://gsnedders.com/>



From sean at miscoranda.com  Tue May 26 10:30:28 2009
From: sean at miscoranda.com (Sean B. Palmer)
Date: Tue, 26 May 2009 18:30:28 +0100
Subject: [html5] Absolute Pixels in HTML 5
In-Reply-To: <F6A515C3-1284-46A6-A592-CBE3615E665D@googlemail.com>
References: <b6bb4d890905250924n27b1d457m4cda9418614b837b@mail.gmail.com>
	<F6A515C3-1284-46A6-A592-CBE3615E665D@googlemail.com>
Message-ID: <b6bb4d890905261030j737cf228t113fdd9b78e8489@mail.gmail.com>

On Mon, May 25, 2009 at 6:45 PM, Geoffrey Sneddon wrote:

> What HTML 5 defines is what browsers currently do, but it's also true that
> all desktop browsers set 1 CSS pixel = 1 physical pixel, so the issue
> doesn't really come up, though the definition matches what happens in cases
> where 1 CSS pixel does not equal 1 physical pixel.

?

-- 
Sean B. Palmer, http://inamidst.com/sbp/


From stevelist at silverorange.com  Fri May  1 06:44:10 2009
From: stevelist at silverorange.com (Steven Garrity)
Date: Fri, 01 May 2009 10:44:10 -0300
Subject: [html5] <video> tag fallback
In-Reply-To: <06DD1FDB-82CC-49B7-8E0D-60B62CA459BD@apple.com>
References: <49E47F2F.2030709@silverorange.com>
	<0FD246DD-3C30-40D8-89F8-7FFF75C3A257@iki.fi>
	<49E4CA2F.7000800@ucsc.edu>
	<DD0C1FCA-09A0-4FCB-AE27-68295F0CDDD9@apple.com>
	<Pine.LNX.4.62.0904300616260.10370@hixie.dreamhostps.com>
	<06DD1FDB-82CC-49B7-8E0D-60B62CA459BD@apple.com>
Message-ID: <49FAFCAA.6040405@silverorange.com>

Eric Carlson wrote:
>   WebKit nightly builds do fire an 'error' event if no suitable src is 
> found.

That's encouraging, but it seems that existing releases of Safari 3.x 
already attempt to support the <video> tag without such an error event. 
So, implementing for real users is already difficult.

If we want a solution that isn't broken for existing Safari users, we'll 
probably have to resort to a browser-detection hack.

Also, throwing a JS exception isn't helpful for users without JavaScript 
enabled.

Not trying to sound like a complainer - just want to make sure you're 
aware of what actual implementers are running into.

Thanks,
Steven Garrity


From caoyp at cn.fujitsu.com  Tue May 19 18:27:24 2009
From: caoyp at cn.fujitsu.com (caoyipeng)
Date: Wed, 20 May 2009 09:27:24 +0800
Subject: [html5] a question about canvas element's shadow effect
Message-ID: <2E9A38A0754E4E22B85E659E028883ED@FNSTcyp>

canvas element's shadow effect use Gaussian blur to fulfill.

In Gaussian blur, when do convolution operation, Gaussian template's size(or
Gaussian kernel width and height) affect the blur effect

,is there default value about Gaussian template's size(such as the size is
(6?+1)*(6?+1),? is the standard deviation).

 

Regards
C.Y.P.

 

Cao Yipeng
Development Dept.I
Nanjing Fujitsu Nanda Software Tech. Co., Ltd.(FNST)
8/F., Civil Defense Building, No.189 Guangzhou Road ,
Nanjing , 210029, China 
TEL: +86+25-86630566-817
COINS: 79955-817
FAX: +86+25-83317685
MAIL: caoyp at cn.fujitsu.com

 

 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.whatwg.org/pipermail/help-whatwg.org/attachments/20090520/5479f90b/attachment-0002.htm>

From sean at miscoranda.com  Mon May 25 09:14:39 2009
From: sean at miscoranda.com (Sean B. Palmer)
Date: Mon, 25 May 2009 17:14:39 +0100
Subject: [html5] Microdata and Meta
Message-ID: <b6bb4d890905250914x41f3c2a1q12a5d58cd89a22ad@mail.gmail.com>

I know the Microdata stuff isn't stable yet, but I was just wondering
if the following is a valid and sensible thing to do:

<!DOCTYPE html>
<head item>
<title>Microdata Example</title>
<link itemprop="about" href="">
<meta itemprop="example" content="value">

The idea is that meta/@itemprop is somewhat like a user space version
of meta/@name, but to enable the former to work exactly like the
latter you have to make it clear that the item being referred to is
the current document. Hence the <link> element in the recipe above.

If that's valid, I wonder if it'd be worth making the following equivalent:

<!DOCTYPE html>
<title>Microdata Example</title>
<meta itemprop="example" content="value">

Mainly because when you use <meta> in head, you already expect it to
be giving metadata about the present document. The item Microdata
construct is basically extending the semantics of <meta>, and when you
use it in its old capacity I wonder if people will get confused. Maybe
not, and they should read the specifications and tutorials properly
anyway; but this is the kind of common sense way I'd expect it to
work.

Also, I must admit that I am prone to making up my own <meta> names
and values. Most of this stuff is half way between being private and
public data, and definitely aligns to the traditional semantics of
<meta> enough to make an @data-* attribute seem like the wrong thing
to do. But I don't know how universal these use cases are.

-- 
Sean B. Palmer, http://inamidst.com/sbp/


From sean at miscoranda.com  Mon May 25 09:24:15 2009
From: sean at miscoranda.com (Sean B. Palmer)
Date: Mon, 25 May 2009 17:24:15 +0100
Subject: [html5] Absolute Pixels in HTML 5
Message-ID: <b6bb4d890905250924n27b1d457m4cda9418614b837b@mail.gmail.com>

HTML 5 currently says that img/@width and so forth use CSS 2.1 pixel
values. Pixel values in CSS 2.1 are of course a relative rather than
an absolute unit. In other words, CSS px is a non-linear alias for
radians; 675.522px, for example, is a quarter of a radian.

But non-vector image formats like JPG and PNG have discrete pixels,
and screens have pixel displays with certain ppi values. So has
anybody looked at whether browsers actually implement px as a relative
unit? And how do they render images? I'm presuming that they don't
resample to fit the relative measurement.

>From the point of view of an author, I'm mainly concerned with, for
example, setting widths for figures. If I'm not giving any height or
width attributes for an <img> and my browser is rendering the image
absolutely, I don't then want to set a relative width for the figure.

-- 
Sean B. Palmer, http://inamidst.com/sbp/


From foolistbar at googlemail.com  Mon May 25 10:45:03 2009
From: foolistbar at googlemail.com (Geoffrey Sneddon)
Date: Mon, 25 May 2009 18:45:03 +0100
Subject: [html5] Absolute Pixels in HTML 5
In-Reply-To: <b6bb4d890905250924n27b1d457m4cda9418614b837b@mail.gmail.com>
References: <b6bb4d890905250924n27b1d457m4cda9418614b837b@mail.gmail.com>
Message-ID: <F6A515C3-1284-46A6-A592-CBE3615E665D@googlemail.com>


On 25 May 2009, at 17:24, Sean B. Palmer wrote:

> HTML 5 currently says that img/@width and so forth use CSS 2.1 pixel
> values. Pixel values in CSS 2.1 are of course a relative rather than
> an absolute unit. In other words, CSS px is a non-linear alias for
> radians; 675.522px, for example, is a quarter of a radian.
>
> But non-vector image formats like JPG and PNG have discrete pixels,
> and screens have pixel displays with certain ppi values. So has
> anybody looked at whether browsers actually implement px as a relative
> unit? And how do they render images? I'm presuming that they don't
> resample to fit the relative measurement.
>
>>
> From the point of view of an author, I'm mainly concerned with, for
> example, setting widths for figures. If I'm not giving any height or
> width attributes for an <img> and my browser is rendering the image
> absolutely, I don't then want to set a relative width for the figure.

What HTML 5 defines is what browsers currently do, but it's also true  
that all desktop browsers set 1 CSS pixel = 1 physical pixel, so the  
issue doesn't really come up, though the definition matches what  
happens in cases where 1 CSS pixel does not equal 1 physical pixel.


--
Geoffrey Sneddon
<http://gsnedders.com/>



From sean at miscoranda.com  Tue May 26 10:30:28 2009
From: sean at miscoranda.com (Sean B. Palmer)
Date: Tue, 26 May 2009 18:30:28 +0100
Subject: [html5] Absolute Pixels in HTML 5
In-Reply-To: <F6A515C3-1284-46A6-A592-CBE3615E665D@googlemail.com>
References: <b6bb4d890905250924n27b1d457m4cda9418614b837b@mail.gmail.com>
	<F6A515C3-1284-46A6-A592-CBE3615E665D@googlemail.com>
Message-ID: <b6bb4d890905261030j737cf228t113fdd9b78e8489@mail.gmail.com>

On Mon, May 25, 2009 at 6:45 PM, Geoffrey Sneddon wrote:

> What HTML 5 defines is what browsers currently do, but it's also true that
> all desktop browsers set 1 CSS pixel = 1 physical pixel, so the issue
> doesn't really come up, though the definition matches what happens in cases
> where 1 CSS pixel does not equal 1 physical pixel.

?

-- 
Sean B. Palmer, http://inamidst.com/sbp/


