Hi José,
Yesterday, you identified a set of functions from GNU libunistring that would
be useful to use in GNU poke. Since you will need only a few such functions,
which sums up to little code and only one big table, you can take the
respective modules from gnulib - a regular use of gnulib-tool. A
Hi José,
you could look at libidn2 as an example how to use system libunistring
if there (or if new enough) and fallback to gnulib unistring.
(BTW, libunistring is made of the gnulib unistring modules)
It creates a separate dir / library for gnulib unistring functions,
*BUT* only uses it when a s
At the first GNU Hackers Meeting of 2020 I presented the recent and current
work on gnulib. Here are the slides.
haible-gnulib-2020.pdf
Description: Adobe PDF document
Hi Tim,
> you could look at libidn2 as an example how to use system libunistring
> if there (or if new enough) and fallback to gnulib unistring.
> (BTW, libunistring is made of the gnulib unistring modules)
>
> It creates a separate dir / library for gnulib unistring functions,
> *BUT* only uses
Hi Bruno,
On 1/13/20 12:01 PM, Bruno Haible wrote:
> Hi Tim,
>
>> you could look at libidn2 as an example how to use system libunistring
>> if there (or if new enough) and fallback to gnulib unistring.
>> (BTW, libunistring is made of the gnulib unistring modules)
>>
>> It creates a separate dir
Hi Tim,
> > A simpler way to achieve the same thing is to include the gnulib module
> > 'libunistring-optional'. It will use the system libunistring if it
> > exists and is new enough, and otherwise compile the respective modules
> > from source.
>
> Ah, I didn't know that, thanks.
>
> Don't tha
Hi Paul,
> that incorrect line comes because ac_compile_for_check_decl is used
> before it is set. And this occurs because Emacs's configure.ac's first
> use of AC_CHECK_DECL is executed only on alpha platforms (which my
> platform is not), which means the initialization of
> ac_compile_for_ch
On 1/13/20 11:02 AM, Bruno Haible wrote:
I would suggest that
the particular 'case' and 'if'/'test' statements - or even the entire main body
of the configure.ac, from line 130 to line 5588 - gets wrapped in an AC_DEFUN
that gets invoked once.
Thanks for the diagnosis. I came up with a simpler
