Re: [Chicken-hackers] Numbers egg interaction with other compiled code.

[Tony Sidaway]

On 10/24/09, Alex Shinn <address@hidden> wrote:
>
> On the one hand, you lose the immediate error reporting and
> instant diagnosis of the problem (oh, this egg isn't using
> numbers).  Not that there's currently any easy way to _fix_
> that problem apart from installing your own modified version
> of the egg.

Well that's it, really.  I would like to be able to keep two flavors
of, say,  SRFI-60 around, so that if an application needs a version
that understands bignums it can seek and load the bignum version.

Here's another problem.  Recall the fred.scm module I showed you at
the top of this thread.  The other day I looked at the manual and
decided to compile it with a require-extension for numbers.

csc -s -require-extension numbers fred.scm -j fred
csc -s -require-extension numbers fred.import.scm

When I loaded fred.so into csi, numbers.so was also loaded.  So far so good.

I then loaded fred.import.so and executed "(import fred").

But then:

>  (f #x7ffffffff)
Error: (negative?) bad argument type: 34359738367

Without actually editing the source file, how can I tell Chicken that
fred.scm must use the numbers egg?  And does it matter if the source
file I'm compiling doesn't include a "(not usual-integrations)"
declaration?

Presumably I'd want to switch off inline compilation during the build, too.