Re: [docbook-apps] change default HTML encoding to UTF-8

["Leif Halvard Silli" <verktystell@russisk.no>]

Hi Bob. Do the stylesheets output both html 4, html 5, xhtml and xhtml5? 
Or did you conflate html 4 and html 5? See more below.

On 14 Aug 2017, at 18:48, Bob Stayton wrote:

> We have a bug report suggesting that the default output encoding for 
> the DocBook html stylesheet be changed from ISO-8859-1 to UTF-8.

I agree with this bug report. Why? Well, for one thing, you - here - 
talk about "html", and "html" today means "html 5". HTML 5.x recommends 
that documents are authored using UTF-8.

Also, when I look at the link in the forwarded message 
(https://www.oxygenxml.com/forum/viewtopic.php?f=6&t=14812&p=43711#p43711), 
I note that the discussion thread talks about HTML 5. I am not able to 
see that HTML 4 is mentioned at all in that thread.

> Note this only applies to the original HTML 4 output from the "html" 
> directory.


Are you saying that the stylesheet also outputs HTML 5? (Note that I ask 
about "HTML 5" and not about xhtml or xhtml5.)


> The "xhtml" and "xhtml5" outputs already output UTF.


The justification for that ought to be that XML defaults to UTF-8. Xhtml 
and xhtml5 are not 'html'.


> The original HTML 4 standard said ISO-8859-1 was the default encoding, 
> but that UTF-8 would be acceptable.

I am not able to find such statement in the HTMl 4 specification. I 
looked at the one page version: https://www.w3.org/TR/html401/html40.txt

UTF-8 ”took over” as the dominant encoding on the Web long before 
HTML 5 became the official version of HTML.

Technically speaking ISO-8859-1 is STILL the default HTML encoding, from 
user agents’ perspective. It is only from an authoring perspective 
that HTML 5 recommends UTF-8.

DocBook stylesheets is an authoring tool. THere is only one processing 
model for HTML, and that model is defined by the latets HTML spec. Thus 
it should use UTF-8.

At the very least, the DocBook stylesheet should not use the HTML 4 
specification as a justification for failing to output HTML 5 as UTF-8.

> It isn't difficult for a user to change the output to UTF-8, but it 
> does require a customization.  The question here is whether to change 
> the default output encoding to UTF-8.

If the user has to change the output to UTF-8 in order to produce HTML 5 
output, then the stylesheet does not follow HTML5’s recommendations.

The fact that the user can produce XHTMl - and thus automatically get 
UTF-8 - does not alter the picture.

> This would change the HTML output to replace character references like 
> &#xXXXX; to actual UTF-8 encoded characters, and change the encoding 
> information in the header to reflect that.

This would be nice. But just for the record: HTML 5.x does not recommend 
against using character references. Thus, if need be, you CAN pick a 
compromise: you can continue to output the character references and yet 
label the document as <meta content="http-equiv" 
content="text/html;charset=UTF-8">. This would then meet HTML 5’s 
recommendation.

> I'm reluctant to change something that will break the builds that 
> DocBook people depend on.  Would this impact you if the change was 
> made?

One thing to perhaps consider s whether interaction between external CSS 
stylesheets (that DocBook may produce) and the HTML output is affected. 
I do not think so, but perhaps there are some edge cases. If you need, I 
can look into it.

Leif

> Bob Stayton
>
>
>
>
> -------- Forwarded Message --------
>
> [bugs:#1400] Default encoding for HTML-based outputs
> .
> Status: open
> Group: output: HTML
> Created: Thu Aug 10, 2017 11:41 AM UTC by Radu Coravu
> Last Updated: Thu Aug 10, 2017 11:41 AM UTC
> Owner: nobody
>
> One of our clients reported that the default output encoding for 
> Docbook to HTML is ISO 8859-1 which is not suitable at all for other 
> languages with extended char sets like Russian:
>
> https://www.oxygenxml.com/forum/viewtopic.php?f=6&t=14812&p=43711#p43711
>
> Maybe the default language for HTML (and also for HTML chunk) should 
> be changed to be UTF-8 as UTF-8 is already used as the default 
> language for XHTML.
>
>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: docbook-apps-unsubscribe@lists.oasis-open.org
> For additional commands, e-mail: 
> docbook-apps-help@lists.oasis-open.org