Re: Opening a can of worms: a readline gnulib module?

[Simon Josefsson]

Stepan Kasal <address@hidden> writes:

> Hello,
>
> On Sat, Jul 16, 2005 at 10:53:45AM +0200, Simon Josefsson wrote:
>> All of this is hypothetical though, since it doesn't look like adding
>> the readline library as a gnulib module will happen now.
>
> Why not?  I don't see any reason against the "readline stub" module
> you proposed.

Neither do I, but with the "readline library" I meant the full 20kloc
of the real readline distribution.  I don't see that part happening
now.  I think it might be useful to turn the readline library into a
gnulib module, but this is more than I need now, and I expect it would
be quite some work involved.

> Bruno pointed out problems with "full readline" module, and asked:
>> > rl_add_defun etc.). Do you want to provide all these symbols?
>
> Your answer was no.
>
> So with your module, the application would have:
> 1) readline() function
> 2) #if READLINE
>
> I guess it might be worth it.  Opinions?

I don't think applications should do #2.  My application only need the
readline interface.  If my application needed more than that, it would
be ugly to start #if-protect the part of the application that the
readline stub didn't implement.  It might as well require the readline
library to be installed.  Or the maintainer who need that
functionality could port the readline library into a gnulib module
proper.

Thanks,
Simon