"Alfred M. Szmidt" <[EMAIL PROTECTED]> writes:

>    > Even if we use the readline module from gnulib, we have to
>    > envelop the add_history calls with #if HAVE_LIBREADLINE...#endif
>    > pairs.
>
>    Yes, or perhaps cleaner: add a AC_CHECK_FUNC(add_history) to
>    configure.ac, and make sure the test code links to any libreadline
>    found by gnulib.  Then wrap the calls to add_history around #if
>    HAVE_ADD_HISTORY.
>
> Nitpick, that should be HAVE_LIBHISTORY, there is no guarantee that
> add_history is part of libreadline.  And the check should check for
> add_history in libreadline _and_ libhistory.

Ah, right.  I forgot that add_history is in libhistory.

>    Another option would be to add the entire libreadline to the gnulib
>    module, so that it really does provide a add_history replacement as
>    well.  But the libreadline code was rather ugly last time I
>    checked, so it may be difficult.
>
> Sounds like a terrible idea to put non-trivial code into gnulib, it
> isn't supposed to be a dumping group for everything in the GNU
> project...

gnulib already contains some non-trivial code like argp, *printf, regex.
Anyway, I guess it is up to the gnulib maintainers to decide.  A
question for InetUtils would be whether libreadline should be a hard or
soft dependency.

/Simon


_______________________________________________
bug-inetutils mailing list
[email protected]
http://lists.gnu.org/mailman/listinfo/bug-inetutils

Reply via email to