Re: [MIT-Scheme-devel] new microcode identification slots for C back end

[Chris Hanson]

Taylor R Campbell wrote:
> Someone recently (well, a month ago) asked me to add a microcode
> identification slot for the type of compiler back end used (i.e.
> whether it is none, native, or C), which sounds reasonable.  I'm also
> considering adding another couple of slots for the C compiler options
> the C linker options, so that we could isolate their detection to the
> microcode's configuration script and remove any hard-coded,
> OS-dependent tables from compiler/machines/C/ctop.scm.  On the other
> hand, it would make the microcode binaries undistributable to systems
> with different C compilers or C linkers from those on the system where
> the microcode was built; however, the issue would still be somewhat
> alleviated by the flexibility in configuration detection instead of
> having a hard-coded table.
> 
> Thoughts?

Why not have the configuration script generate shell scripts with the
appropriate calling sequences, and install them in lib/mit-scheme?  That
would have roughly the same effect and also allow overriding by the end
user.