[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Arch-independent parts of binary modules of interpreted languages



(Please don't start a debate over what an interpreted language is, I just 
tried to generalize the subject.)

Perl XS module packages usually install all their code under /usr/lib/perl5 - 
not just the shared library that implements the external subroutine, but also 
at least one ordinary module, which interfaces with the shared library using 
DynaLoader and perhaps provides additional subroutines. These ordinary 
modules are not architecture-specific in themselves. I don't know that much 
about Python - IIUC the interpreter can directly load shared libraries that 
implement the right interface, but python2.4 and python2.5 at least 
install .py files in /usr/lib/python<version>.

What's the rationale behind not strictly separating architecture-independent 
and architecture-specific code? I'm trying to find out if I can apply the 
same rationale to the pike packages, which I'm adopting. There the files are 
separated, with symlinks from /usr/lib/pike<version> to the corresponding 
location under /usr/share/pike<version>.

-- 
Magnus Holmgren        holmgren@lysator.liu.se
                       (No Cc of list mail needed, thanks)

  "Exim is better at being younger, whereas sendmail is better for 
   Scrabble (50 point bonus for clearing your rack)" -- Dave Evans

Attachment: pgpdXyfzJxef7.pgp
Description: PGP signature


Reply to: