[Pyrex] ANN: Pyximport

[Paul Prescod]

Arnd Zapletal wrote:

> Paul Prescod <paul at prescod.net> wrote:
> 
>  >>http://www.prescod.net/pyximport/
> 
> Paul,
> 
> finally found some time and had a try:
> I believe there is a minor mess with your module names (probably you renamed a former version?)

You're right. I slip-streamed a fix for that (the "1.0" on the Web now 
has the fix) and I'm going to announce a 1.1 on the weekend with that 
fix, another bug fix and two new features:

  * ability to specify dependencies
  * ability to pass parameters to distutils like other ".c" files to 
compile or build options, etc.

One thing I'm going to have to spend a little time on is getting 
distutils to give me a little more control where the "build" directories 
go. I think right now it is based on you getcwd() whereas I'd like it 
always to be relative to the .py file.

> I changed this, *afterwards* the test-run worked and -Hooray- I can run my stuff
> directly from SciTE now. Nice. 

AWESOME.

> Hmm. I wonder whether pyximport could be integrated into the pyrex-package? 
> So one could pass source code directly to others, and non-pyrexers/-pythonics wouldn't have to care about compiling it (explicitely).
> 
> This was your idea, or?

Eventually. I'd give Pyximport a few months to shake out bugs before I 
proposed it to Greg but yes, I think that the implicit compile way of 
doing things is fundamentally better (at least as an option). Even 
luddites like Sun and Microsoft use this technique for ASP and JSP. But 
I still think Python was the pioneer. I'd be curious to know of 
precedent for the way Python handles .py and .pyc files.

  Paul Prescod