[pypy-dev] Separate building of the C source files

Elad Lahav e2lahav at gmail.com
Wed Sep 14 16:12:47 CEST 2011


I am trying to build Pypy for an embedded platform, with its own build
system. For that purpose, I would like to have the translation process run
on a Linux/x86 box, generating the C source files, and then have the
platform's own build system do C compilation and linking. With the 1.6
sources, I used the following command:

$ python translate.py --opt-jit --source targetpypystandalone.py

The source files were created under the /tmp directory. The first thing I am
missing, though, is a makefile. Looking at the translator sources, it
appears as though a GNU make-compatible makefile should have been created,
but I cannot find one. Is such a makefile written to disk as part of the
source generation stage, or does the translator use a different mechanism?

Instead of using a makefile, I also tried manual compilation of the files
under /tmp/usession-unknown-NUMBER/testing_1. The linking stage (with -lffi
-lm -ldl -lpthread) fails with unresolved symbols, which I do not recognize
from any of the common libraries. Nor could I find any documentation of the
post-translation build process. Can anyone describe the C compilation and
linking commands, or point me to the right place in the code/documentation?

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/pypy-dev/attachments/20110914/18278db5/attachment.html>

More information about the pypy-dev mailing list