Re: Add support for generating HTML docs a` la PDF, etc., revision 2

[Alexandre Duret-Lutz]

>>> "Ralf" == Ralf Corsepius <address@hidden> writes:

 Ralf> Am Fre, 2003-02-28 um 12.30 schrieb Richard Dawe:
 >> Hello.
 >> 
 >> Below is a revised version of the patch to add support for an 'html'
 >> target for generating HTML using makeinfo. Changes:

 Ralf> Hmm, this just caught my eye:
 
 >> -.PHONY: info info-am dvi dvi-am pdf pdf-am
 >> +.PHONY: info info-am dvi dvi-am pdf pdf-am html html-am
 Ralf> Is it really a good idea to make "info", "dvi", "pdf" and "html"
 Ralf> make-targets PHONY?

 Ralf> I think, there non-negligible possibility for them to conflict with
 Ralf> sub-directories, eg. if docs are generated dynamically (eg. doxygen
 Ralf> generating its docs into html/) or with subdirectories containing
 Ralf> include-files for documents (e.g. texi-docs conditionally including
 Ralf> files by @ifhtml/@ifinfo etc.).

 Ralf> Esp. html seem critical to me, because unlike dvi, pdf
 Ralf> and ps, html docs often are split into several
 Ralf> sub-documents.

I'm a bit confused by this.  Could you provide an scenario for
such a situation (i.e., where `.PHONY: html' is harmful)?
AFAICT, we need `.PHONY: html' precisely to work around cases
where `html/' exists.  Without PHONY, `make html' would stop its
recursion as soon as it encounters a `html' file or directory.

Maybe you are talking about a user-supplied `html:' rule that
build a `html/' directory?  Still I don't really see how
`.PHONY: html' would harm.  Are there cases where you type `make
html' and don't want the documentation to be rebuilt?  (E.g.,
with Doxygen I'd assume that short of accurate dependencies you 
always want to rebuild the doc.)
-- 
Alexandre Duret-Lutz