Re: with HTML output, @minus{} is converted to a hyphen instead of a real minus character

[Patrice Dumas]

On Wed, Oct 12, 2022 at 01:59:12PM +0200, Vincent Lefevre wrote:
> On 2022-10-12 11:11:53 +0200, Patrice Dumas wrote:
> > On Wed, Oct 12, 2022 at 01:13:07AM +0200, Vincent Lefevre wrote:
> > > With Texinfo 6.8 and HTML output, @minus{} is converted to a hyphen
> > > instead of a real minus character (U+2212 MINUS SIGN).
> > 
> > It is actually not clear to me why an ASCII - is output in the default
> > case, and not −.  Maybe it was for compatibility with the US-ASCII
> > default encoding?
> 
> This should normally be unrelated: the encoding is about the HTML
> source, while the "output charset" has been Unicode for a long time,
> and there are no reasons to have the same requirements about the
> encoding. Character references such as "−" were introduced
> in HTML for this reason.

I remember some time ago, probably when latin1 was the default charset,
that there could be some entities not formatted.  But it was probably
some time ago.  Note that we want to output HTML that is ok on old
browser, so we have to be conservative.  We are outputting entities out
of the charset for some time, though, so this is probably not a concern
we should have.

> > This could be changed now the UTF-8 is the default encoding.
> 
> Is it? I cannot see any change in the NEWS file in master.

You are right, I will add it.

> > Opinions?
> 
> I agree that UTF-8 should be the default encoding. Then the behavior
> with old encodings would be less important.

I was asking the opinions on using − instead of -, not on UTF-8
being the default encoding, which is effective now...

> The NEWS file in master currently says:
> 
>  . HTML/DocBook output:
>      . meaning of USE_NUMERIC_ENTITY changed to be using numeric entities in
>        priority.  former effect is now always on.
>      . ENABLE_ENCODING_USE_ENTITY variable removed.  former effect is now
>        always off.
> 
> I wonder what this means, in particular if US-ASCII is used.

It changes nothing in the default case, only if the USE_NUMERIC_ENTITY
customization variable was used.  Even in that case, the previous
behaviour probably was not what the user expected.  My wild guess is
that no user is affected, but it was used in the test suite.

-- 
Pat