[Pharo-project] Popularity of Smalltalk in Software Industry
siguctua at gmail.com
Fri May 6 15:08:50 CEST 2011
On 6 May 2011 10:30, Stéphane Ducasse <stephane.ducasse at inria.fr> wrote:
> We should face it: the FFI and Alien... are not that simple. Laurent spent time to try to cnnect to libs and it did
> not work. So we should really improve that aspect.
> Igor not everbody is able to code in assembler for nativeboost.
I am not presumed that everyone should.
But just don't fall into delusion, that you can start using external
library (either written in C or C# or in Java),
without knowing at all the foreign language in which library are
written, and just after few mouse clicks in Pharo image.
The library are written not in smalltalk, and if you need to use it,
you have to be prepared.
You cannot avoid that. That's my point.
> So NB is probably a good insfrastructure but not ready for consumption for everybody.
Mind telling me, what is missing?
I am eager for improving and making it easy to use.
> The interaction with C should be improved :)
How? Any clues?
How else it could be improved in addition to be able to define, which
function you going to call and using which library?
How else it could be done?
Parsing C headers and automatically generating bindings? This is
doable. But it still will require manual effort.
You cannot build an auto-generator to be intelligent enough to detect
all specific aspects, which used in library.
Read a http://www.swig.org/
they having a plugin system (and at some point i even written a
smalltalk plugin for it).
But even then, for generating a specific library bindings, you still
have to provide a script/configuration for your library.
Because if you run it in automatic mode, in most cases you'll end up
with a lot of useless and/or broken stuff.
Translating (producing wrappers) from one language into another
requires intelligent effort,
it cannot be fully automated.
Igor Stasenko AKA sig.
More information about the Pharo-project