Re: Macro argument quoting and ifelse condtional

[Gary V. Vaughan]

Bill McCarty wrote:
> Hi all,
>
> I'm new to m4 and finding an apparently common idiom
> confusing. Can someone help me understand it?
>
> Assuming quotes have been replaced by << and >>, a
> typical macro definition of the sort that puzzles me
> reads as follows:
>
> define(<<_CF_OUTPUT>>, <<dnl
> ifelse(_CF_$1, <<_CF_$1>>, <<>>, <<
> divert
> CF_$1_OUTPUT
> divert(-1)
>
> > > )>>)dnl
>
> dnl
>
> I'm particularly puzzled by the distinction between
> the first two arguments of ifelse: _CF_$1 and
> <<_CF_$1>>. The ifelse seems to cause the macro to
> expand to void unless its argument is double-quouted,
> like <<<<TEST>>>>. But, in actual use contexts, its
> argument seems generally to be single-quoted. So,
> genius that I am <g>, I conclude that I'm missing
> something. But, I can't see what.

_CF_OUTPUT(test)
  -> ifelse(_CF_test, <<_CF_test>>, <<>>, ...

as each arg is parsed one level of quoting is removed, as
each is evaluated another level of quoting is removed.  So
we eventually compare:
    _CF_test and _CF_test

hence the 3rd arg is evaluated.

Similarly for _CF_OUTPUT(<<test>>)
  -> ifelse(_CF_<<test>>, <<_CF_test>>, <<>>, ...


However:

_CF_OUTPUT(<<<<test>>>>)
  -> ifelse(_CF_<<<<test>>>>, <<_CF_<<<<test>>>>, <<>>, ...

2 levels of quotes are removed before the comparison, hence:

  _CF_test and __CF_<<test>>

are compared, so the 4th argument of ifelse is evaluated.

HTH,
        Gary.
--
Gary V. Vaughan      ())_.  address@hidden,gnu.org}
Research Scientist   ( '/   http://tkd.kicks-ass.net
GNU Hacker           / )=   http://www.gnu.org/software/libtool
Technical Author   `(_~)_   http://sources.redhat.com/autobook

signature.asc
Description: OpenPGP digital signature