On Jul 6, 2007, at 7:37 PM, Hazen Babcock wrote:
> On Jul 6, 2007, at 8:27 AM, Jerry wrote:
>> I hope there is another Mac user on the list.
>> For a while, my installation has been such that I can configure
>> PLplot for either Ada bindings or to output to AquaTerm (aqt) but not
>> both. The Ada compiler that I've been using from macada.org, version
>> 4.2, doesn't have Objective C in it, and PLplot needs Objective C to
>> make the AquaTerm driver.
I should have mentioned that this problem first appeared when I
started downloading from CVS a few months ago and using the developer
build tools provided with PLplot. Before that, I just built PLplot
like an ordinary user who had downloaded the latest stable and then
made sure that my tools knew where to link to--that after downloading
Aquaterm and doing the same with it. I'm sure that I could do that
once more but that would take away from being able to work from SVN.
> Do you think it is possible to use one compiler for Ada and another
> for the rest of PLplot, sort of like what is done with Fortran?
Yes. That is what I had in mind when I mentioned using the 4.2
version of the Ada compiler (which is not at /usr/bin and which
includes C but not Objective C), as I have been doing recently, and
then telling Cmake where to find the Apple C/C++/Objective C/
Objective C++ compiler (which is at /usr/bin). Currently, Cmake is
told only where the Ada/C compiler is, then (I suppose) gives up when
it discovers that there is no Objective C compiler there. Presumably
something similar would happen if I enabled the C++ bindings while
using the 4.2 compiler.
>> (For what it's worth, Cmake doesn't appear to test for the presence
>> of an Objective C compiler even though Aquaterm needs it. The
>> Aquaterm framework is found, however.)
> The assumption is that people who have AquaTerm compiled and
> installed it using Apples developer kit. I don't think there is a
> AquaTerm binary so this seems reasonably safe.
I think that has been a good assumption. However, with Ada in the
picture, Cmake needs to know where to find the Ada compiler as a
separate step from knowing where to find the Apple Objective C
compiler. This is why I thought I could just put the new 4.3 compiler
which contains Ada, C, and Objective C (and others) in the location
where we are telling Cmake to find "the compiler." I don't know why
that didn't work.
So, in part, it appears to me, Cmake needs to be told (1) the
location of the Ada compiler (which we currently do, at /usr/local/
whatever), and (2) the location of the Objective C compiler at /usr/bin.
-- OR --
Figure out why the more-complete set of compilers that I have called
the 4.3 compilers failed when being asked to work with Objective C.
To tell you the truth, I really don't know why Object C is needed,
since Aquaterm is already compiled.
And all of this with the caveat that I really don't understand much
of the machinery of which I am writing about. Maybe I am completely
misinterpreting the error output that I included in my original post.
FWIW, I think that there's a chance that Apple will include the Ada
compiler as an official part of the release due in October 2007.