Solaris 64 bit compilation

[Martin v. Loewis]

phoebe_1 at att.net (Holden Caulfield) writes:

>  This is my first attempt at compiling Python v2.2.1 in *64bit* mode 
> on a Solaris 5.9 system. I am using the Sun Compiler (the latest I do not
> remember the version). I am migrating a few of my apps (all developed on
> 64bit platforms) to Solaris, so I *need* the 64bit mode.


I can't follow your conclusion - if you migrate applications which
have been developed on 64-bit platforms, there is no inherent need to
compile them as 64-bit binaries on Solaris 9.

> But, it is failing miserably. It compiles *fine* in 32bit mode (the default),
> but it does me no good. As some of the extensions library that I have are
> 64bit. Before, I start hacking I thought I would give a query here and see
> if anyone had any success?

It compiles fine for me, with SunPRO 6U1.

> 
> Here are my preliminary results:
> - Flags used "-xarch=v9 -xcode=pic32".
> The first thing I get is the bail-out in "pyport.h":
>  Where the #LONG_BITS != (8 * sizeof(LONG)) fails!!!

Please try to report failures carefully. It does not say sizeof(LONG),
instead, it says SIZEOF_LONG. Which gives rise to the question: What
value does SIZEOF_LONG have, in pyconfig.h?

> So, I commented it out, first to proceed and see what happens. Most of them
> compiled fine except, I got some serious errors in the:
> fileobject.c file probably related to the LP64 model.
> 
> I will try to add more info, when I get around to it.

More info would be required if you want help. Start with reporting the
error messages as the compiler produces them, instead of interpreting
them, and reporting your interpretation.

Regards,
Martin