Re: Re-write of documentation tree build process

[John Mandereau]

Jan Nieuwenhuizen wrote:
> John Mandereau <address@hidden> writes:
> > If I really wanted to rewrite it, I'd split it into:
> > - a function that walks the input directories and returns directories
> > and files lists
> > - filtering these lists
> > - mapping the lists with mkdir, hardlink and symlink standard library
> > functions
> 
> Yes, that makes sense.

In mid-January, I'll have some time to work on this (and on the plan
below).



> As Han-Wen says, the millions of options is hardly a good thing about
> rsinc.  Mix that with the fact that the ordering of some options is
> significant, rsync has intricate special regular expression syntax
> '**' comes to mind, and trailing slashes sometimes have magic meaning,
> all makes it next to impossible to predict what rsync will do, without
> actually testing it.

You're definitely right: I had to test 'rsync -rl' in
GNUmakefile.in(install-WWW) to get the desired result, with a trailing
slash for the source and no trailing slash for the destination.

I think add_html_footer.py could be even more cleaned up this way:

- drop do_file and its ton of arguments, move the file I/O code to the
loop on the file list in add_html_footer(), and move text substitution
code to small functions. As a consequence, all this mess would be easier
to maintain: disabling a text substitution would mean commenting a
single function call line, and the order of text substitutions would be
clearer (target-specific or not, pre- or post-.html link
processing, ...)

- decrease the number of arguments passed to add_html_footer() by
setting as much as possible global variables in add_html_footer.py, just
like current default_footer.

When this is done, it'll be possible to i18n the footer and tha language
menu without making this script a mess again. Any thoughts?

-- 
John Mandereau <address@hidden>