Re: py_cpp and Compaq Alpha Tru64

From: David Abrahams (abrahams_at_[hidden])
Date: 2000-10-13 07:07:56

----- Original Message -----
From: "Ralf W. Grosse-Kunstleve" <rwgk_at_[hidden]>
To: <abrahams_at_[hidden]>
Cc: <rwgk_at_[hidden]>
Sent: Friday, October 13, 2000 3:01 AM
Subject: py_cpp and Compaq Alpha Tru64

> I am trying to compile py_cpp with the native cxx compiler on a Compaq
> Alpha running Tru64 5.0. With some effort (I am a C++ beginner) I
> figured out how to install STLport-4.1b3. The STLport regression test
> checks out OK.
>
> The py_cpp files I am trying to compile are:
>  extclass.cpp functions.cpp init_function.cpp module.cpp newtypes.cpp
>  objects.cpp py.cpp subclass.cpp
>
> For each file there are warnings about long long being nonstandard.
> I assume these warning can safely be ignored.

I think so, and they're produced by the Python headers. I would recommend
disabling any warning which you plan to routinely ignore.

> However, for one file,
> newtypes.cpp, a long list of error messages is produced. The cxx
> command I used and the full output are attached.
>
> Are the errors due to a limitation of the cxx compiler?

Well, it's hard to say. I know absolutely nothing about this compiler. I
think that most compiler front-ends for workstations such as Alpha were
written by EDG, which means that if the compiler is reasonably up-to-date,
it is highly conformant. I have no idea, though, what version of this
compiler you have nor what version of EDG's front-end it would use, so I
can't even speculate.

> Would you have ideas for fixes or work-arounds?

One possibility is that the compiler is getting confused by the
Windows-style CR/LF line endings in newtypes.h/newtypes.cpp. I notice that
all of the error messages refer to uses of multiline macros. I have posted a
new version with only Unix line-endings. You might try stripping the LFs
from what you have or downloading the new version (that's the only change).
If that doesn't work I suggest preprocessing the file you're trying to
compile and feeding the result to the compiler to get a better sense of
where the error really is.

Good luck, and please let me know how it works out,
Dave