Re: UTF-8 conversion problem in Texinfo 6.6 with TEXINFO_XS_PARSER

[Patrice Dumas]

On Tue, Feb 19, 2019 at 06:22:20PM +0200, Eli Zaretskii wrote:
> > 
> > Looking at the perl parser, it seems that the encoding is reset for all
> > the input files upon reading @documentencoding, so a @documentencoding
> > in an include file resets the main file documentencoding.  I would say
> > that this is incorrect.
> 
> Whether a mistake or not, this command exists in Texinfo for too long
> to change it now in backward-incompatible ways.  It always produced a
> global setting which was in effect for the whole document.
> 
> The main use case of @documentencoding is exactly to specify a single
> encoding for the entire document.  The use case you propose sounds
> like a much more marginal one.  In fact, I cannot even envision when
> would that be useful -- to produce an HTML document where each section
> is in a different language, perhaps?

No, but to handle includes of input files that are not in the same
encoding.

> So I think if we want such a file-local encoding, we should introduce
> a new command, @fileencoding, say.  That would be backward-compatible.

I think that it could be a good idea.  Currently, @documentencoding does
two things, it specifies the input encoding, but also the output
encoding.  I think that it is not a good idea to do both, as the output
encoding could be different from the input encoding and different for
different output formats.  For instance, in XML and Docbook output, UTF-8
is always used for the output.

But even in that case, it could be relevant to change in a not backward
compatible way the meaning of @documentencoding, such that it only
specifies the output encoding.

-- 
Pat