Re: Patch for "HTML5" support

Marc Gueury <[email protected]> Mon, 07 Nov 2011 11:07:46 +0100
Newsgroups gmane.comp.web.html-tidy.devel
Message-ID <[email protected]>
This is a multi-part message in MIME format.
--===============0284707198772187820==
Content-Type: multipart/alternative;
	boundary="------------090701020601090504030807"

This is a multi-part message in MIME format.
--------------090701020601090504030807
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit

Hi Bjoern,

I have seen no answer for this good idea yet.
I do not know exactly what to think about it and I am probably not the 
best person to answer this.

The real problem is that Tidy should not create a file with more error 
after cleanup of the HTML than before
cleanup. If it is the case for nearly all the cases, then it is very nice !

_/A. Commit the patch or not/_

- if such change does not happen, I think that we can say goodbye to 
Tidy in the next years when all is HTML5.
- I think that there is no replacement about a feature like Tidy in 
HTML5 for now.

So, for these reasons alone, I think it is a good move. Your code should 
be checked in and
tested with real pages to check. Even if the program is not perfect yet.

_/B. HTML 5 validator comparison /_

- I have personally currently not a lot of free time and I have never 
read the HTML5 spec fully
- While I understand that there is a parser algorithm part of the HTML5 spec
- I do not understand where is the new "conformance checker" 
specification defined ? I know only that it is a sum of
   - some hand written code (to parse?)
   - schema validation
   - and some XPATH verification of the tree.

Due to the ""some hand written code"", I do not see why to let Tidy live 
in parallel with this ?
Maybe if maybe at a later time the parser part of the code could be 
taken to do some kind of Tidy work ?

_/C. Your last issues
/_I think there is a UTF8 output to Tidy and with it, the entities issue 
is not very relevant.
When the output is ASCII only, I think the entities becomes then relevant.
Does it answer your question ?

Thanks,

Marc



On 05/11/2011 05:51, Bjoern Hoehrmann wrote:
> * Bjoern Hoehrmann wrote:
>> If there is enough interest in this that we get some test reports to the
>> [email protected] mailing list, and people can't find major
>> bugs, I might polish the patch and commit it.
> http://lists.w3.org/Archives/Public/www-archive/2011Nov/0007.html has a
> bit more polished version, I sorted some table entries, this should deal
> with data-* attributes, and I made the tidyAttrIsProp changes I noted. I
> also note that none of the "HTML5" named character references are added.
> I am unsure how the logic for those should work, four problems:
>
>    * It is unclear to me that this new feature has been properly tested
>      in the real world, supporting all the new ones may break things.
>
>    * There are now "foo" and "foo;" entities to accomodate the "missing
>      semicolon" problem to some degree. Tidy could do the same simply by
>      keeping the semicolon around with an appropriately changed table
>
>    * Some entities apparently resolve the multiple characters now. That
>      is a bit of a problem as it makes generating the entities from the
>      characters a good bit harder; also, this would require changes to
>      the data tables, would probably have to write { 0xABCD, 0x0000 } or
>      some such for lack of Unicode escape support in old compilers.
>
>    * Some of these entities are fairly new. When should Tidy generate
>      the new ones? Should there be an option to control this?
>
> Generally, I would probably put support for them on the "someday, maybe"
> pile along with inline SVG and MathML support, when there is evidence of
> these features being supported (by, say, all major search engines, where
> such support matters).


--------------090701020601090504030807
Content-Type: text/html; charset=ISO-8859-1
Content-Transfer-Encoding: 7bit

<html>
  <head>
    <meta content="text/html; charset=ISO-8859-1"
      http-equiv="Content-Type">
  </head>
  <body bgcolor="#FFFFFF" text="#000000">
    Hi Bjoern,<br>
    <br>
    I have seen no answer for this good idea yet. <br>
    I do not know exactly what to think about it and I am probably not
    the best person to answer this.<br>
    <br>
    The real problem is that Tidy should not create a file with more
    error after cleanup of the HTML than before<br>
    cleanup. If it is the case for nearly all the cases, then it is very
    nice !<br>
    <br>
    <u><i>A. Commit the patch or not</i></u><br>
    <br>
    - if such change does not happen, I think that we can say goodbye to
    Tidy in the next years when all is HTML5.<br>
    - I think that there is no replacement about a feature like Tidy in
    HTML5 for now.<br>
    <br>
    So, for these reasons alone, I think it is a good move. Your code
    should be checked in and<br>
    tested with real pages to check. Even if the program is not perfect
    yet.<br>
    <br>
    <u><i>B. HTML 5 validator comparison </i></u><br>
    <br>
    - I have personally currently not a lot of free time and I have
    never read the HTML5 spec fully <br>
    - While I understand that there is a parser algorithm part of the
    HTML5 spec<br>
    - I do not understand where is the new "conformance checker"
    specification defined ? I know only that it is a sum of<br>
    &nbsp; - some hand written code (to parse?)<br>
    &nbsp; - schema validation<br>
    &nbsp; - and some XPATH verification of the tree.<br>
    <br>
    Due to the ""some hand written code"", I do not see why to let Tidy
    live in parallel with this ? <br>
    Maybe if maybe at a later time the parser part of the code could be
    taken to do some kind of Tidy work ?<br>
    <br>
    <u><i>C. Your last issues<br>
      </i></u>I think there is a UTF8 output to Tidy and with it, the
    entities issue is not very relevant.<br>
    When the output is ASCII only, I think the entities becomes then
    relevant.<br>
    Does it answer your question ?<br>
    <br>
    Thanks,<br>
    <br>
    Marc<br>
    <br>
    <br>
    <br>
    On 05/11/2011 05:51, Bjoern Hoehrmann wrote:
    <blockquote
      cite="mid:[email protected]"
      type="cite">
      <pre wrap="">* Bjoern Hoehrmann wrote:
</pre>
      <blockquote type="cite">
        <pre wrap="">If there is enough interest in this that we get some test reports to the
<a class="moz-txt-link-abbreviated" href="mailto:[email protected]">[email protected]</a> mailing list, and people can't find major
bugs, I might polish the patch and commit it.
</pre>
      </blockquote>
      <pre wrap="">
<a class="moz-txt-link-freetext" href="http://lists.w3.org/Archives/Public/www-archive/2011Nov/0007.html">http://lists.w3.org/Archives/Public/www-archive/2011Nov/0007.html</a> has a
bit more polished version, I sorted some table entries, this should deal
with data-* attributes, and I made the tidyAttrIsProp changes I noted. I
also note that none of the "HTML5" named character references are added.
I am unsure how the logic for those should work, four problems:

  * It is unclear to me that this new feature has been properly tested
    in the real world, supporting all the new ones may break things.

  * There are now "foo" and "foo;" entities to accomodate the "missing
    semicolon" problem to some degree. Tidy could do the same simply by
    keeping the semicolon around with an appropriately changed table

  * Some entities apparently resolve the multiple characters now. That
    is a bit of a problem as it makes generating the entities from the
    characters a good bit harder; also, this would require changes to
    the data tables, would probably have to write { 0xABCD, 0x0000 } or
    some such for lack of Unicode escape support in old compilers.

  * Some of these entities are fairly new. When should Tidy generate
    the new ones? Should there be an option to control this?

Generally, I would probably put support for them on the "someday, maybe"
pile along with inline SVG and MathML support, when there is evidence of
these features being supported (by, say, all major search engines, where
such support matters).
</pre>
    </blockquote>
    <br>
  </body>
</html>

--------------090701020601090504030807--


--===============0284707198772187820==
Content-Type: text/plain; charset="us-ascii"
MIME-Version: 1.0
Content-Transfer-Encoding: 7bit
Content-Disposition: inline

------------------------------------------------------------------------------
RSA(R) Conference 2012
Save $700 by Nov 18
Register now
http://p.sf.net/sfu/rsa-sfdev2dev1
--===============0284707198772187820==
Content-Type: text/plain; charset="us-ascii"
MIME-Version: 1.0
Content-Transfer-Encoding: 7bit
Content-Disposition: inline

_______________________________________________
Tidy-develop mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/tidy-develop

--===============0284707198772187820==--