numpy & non-Apple gcc?

[Michael Dickens]

Hi Keith - No problem on the "reply to all" thing; it happens to all of us (really).  There was a follow-up on this subject (renamed "ambivalence about fortran") that covered parts of your fortran question.  Apple does not provide a fortran compiler in XCode, and, really, there is no good way to bootstrap or patch in fortran to that compiler; so if any given port requires fortran for compiling then some MacPorts port needs to be installed to handle it.  The options seem to be "g95" and one of the "gcc4X" series.  Given that there is no way for a port to depend on a specific -variant- of another port (e.g., "gcc +fortran" only assuming that there were variants for the other compilers), one has to select a specific port and use all of it.  For the scientific / math ports that require fortran, the gcc4X series was picked; I was not part of this discussion, so I cannot say the reasoning behind it.  There are some ports that do not require MP's gcc4X's fortran at all (no fortran linking at all), while others such as numpy may or not -require- it but do have variants that use libraries that do require it -- and, hence, should really be using MP's gcc4X in this case to be consistent.  I don't know if the above answered your questions well enough; if not, ask again in another manner. - MLD

On Sep 21, 2010, at 1:00 PM, Keith J. Schultz wrote:
> First off I would like to apoligize for sending mails to the
> original sender and not the list. It is the first time this has happen.
> But, I usually expect that when receiving  mails from a list that the
> return path is set to the list. This seems not to be the case. I will try
> to be more careful from now on.
>
> I should have been more specific in my first message. 
> What I meant was could we get the fortran capabilities or compiler
> patched into Apple's or have a fortran compiler in MacPorts offering
> similar capabilities, so that other ports do not have similar problems.
>
> If would think the only problem is bootstrapping the gfortran compiler.
> But, I do not know the intrinsics of gcc and the gfortran compiler.