Re: MELT tutorial on the wiki

[Dave Korn <dave dot korn dot cygwin at googlemail dot com>]

Basile STARYNKEVITCH wrote:
> Dave Korn wrote:
>> Tom Tromey wrote:
>>
>>> I looked into this a little.  It looks like the PPL checks don't work
>>> properly in the case where PPL is a system library.  I guess I need
>>> --with-ppl=/usr ... I will try that later.
>>
>>   Were you using a --prefix?  The PPL checks (by design I think) only
>> look for PPL in your prefix.
> 
> Sorry DaveK, are you talking of only the MELT branch or of the current
> gcc trunk (future 4.5)? 

  Trunk.

> In the latter (trunk) case, what is the
> rationale for checking only in the prefix?

  I do not know it; I have merely observed the behaviour.  It may even not be by
design for all I know, though I suspect it makes sense - where else would you
look but in the prefix?  Prefixes exist to create separation between packages.

> I have to clean up a bit my MELT's gcc/configure.ac, but I cannot
> understand why apparently the trunk's gcc/configure.ac does not set any
> HAVE_ppl flag.
> 
> Why is there no
> if test "x${PPPLLIBS}" != "x" ; then
>    AC_DEFINE(HAVE_ppl, 1, [Define if PPL is in use.])
> fi
> near line 4110 of the trunk's gcc/configure.ac ?

  I imagine nobody has needed one yet.

> IMHO, such a test and such a generated #define makes a lot of sense (at
> least inside plugins).

  It certainly does.  I'm sure there's no reason not to add it.

    cheers,
      DaveK