On April 7, 2016 5:29:14 PM EDT, "Víctor Cuadrado Juan" <m...@viccuad.me> wrote:
>I have come across an upstream that ships both the cythonized .c file
>and the .py source, on my ITP python-neovim-gui [1].
>
>On #python @freenode I have been said that shipping both files is
>standard practice, which seems to be backed by the python docs [2].
>
>I understand the need to separate “bundled files for end users”, versus
>“actual source release" (aka the need to separate the build step),
>essential to both software freedom and the utilitarian view of working
>with the source (to fix bugs, fork, etc).
>
>I also understand that the Python folks don't want the end user to
>depend on cython and all that means, so they choose the middle ground
>of providing both files.
>
>I haven't found previous talk on this, has this topic already been
>brought to the Cython/Python folks before?

The presence of binary artifacts in the source tarball isn't a problem as long 
as the relevant source in preferred form of modification is also present and 
the binary can be built using tools in Debian.

The only way to reliably know if it can be built is to do so as part of the 
package build, so you should regenerate the binary file and install that in 
your package.

This occurs many places in the archive and isn't at all unique to 
cython/python.  I don't think there's an issue that needs discussion as long as 
the actual (pre-cython) source is included.

Scott K

Reply via email to