[physfs] Wishlist

Ryan C. Gordon icculus at icculus.org
Wed Feb 3 00:06:03 EST 2010


> That sounds not so good. Okay, templates are really problematic, but i
> hoped SWIG could handle them. I better don't ask about boost ;) 

It can, but it needs you to explicitly tell it what to instantiate, and 
how to map it to the scripting language. Since templates are dealt with 
by the C++ compiler, you couldn't have a python script that instantiates 
a template that wasn't already pushed through the static C++ compiler.

(and since MyTemplate<blah,some_expression(),int> doesn't map to most 
scripting languages, you need to tell it you want that to be 
"mytemplate_uniquename" when you reference it in the script.

Most of the problems you'll have with SWIG is that it's not a full C++ 
parser, so if you just point it at some complicated C++ code and tell it 
to figure out the interfaces, it might fail. But even there, you can 
just write your own .i file that doesn't have much C++ in it, or deal 
only with C-callable APIs in your C++ code, or whatever.

> It's my first time with Mercurial and it looks hmm, interesting. Is it
> really easy to merge development branches and keeping them in sync?

Yes.  :)

--ryan.






More information about the physfs mailing list