|
From: Roman Y. <rom...@gm...> - 2007-05-12 17:30:11
|
On 5/12/07, Gustavo Carneiro <gjc...@gm...> wrote: > > On 12/05/07, Roman Yakovenko <rom...@gm...> wrote: > > > > On 5/12/07, Gustavo Carneiro <gjc...@gm...> wrote: > > > Sometimes Py++ _knows_ a method won't compile, but tries to bind it > > anyway. This can be bad when you have developers adding C++ methods and not > > aware or not caring about Python bindings; sometimes they add methods that > > won't compile. > > > > >Not to mention that is a pain to have to manually exclude the methods > > that won't compile. > > > > Why? I think Py++ provide a convenient interface for such things. > > > Yes, Py++ provides the best possible interface for everything. But > sometimes no interface at all is even better ;-) > I agree with you. > Couldn't Py++ just not bind methods that will not compile for sure? > > > > Yes. Today every declaration has "exclude" method. I will add new > > argument - "declarations_that_cause_compile_time_errors_only" ( please give > > a better name ). > > You will have to set it to "True". The function will write to the log > > all declarations it excluded. > > > > I guess you don't want to exclude function that missing call policies, > > right? > > > Yes, missing call policies is what usually hits me. I am not sure if > there are other cases when Py++ definitely knows the generated code won't > compile... > Yes it knows: http://pygccxml.svn.sourceforge.net/viewvc/pygccxml/pyplusplus_dev/pyplusplus/messages/warnings_.py?view=markupsome of the warnings related to functions. Also you don't have to wait for the patch - you don't need one mb = module_builder_t(...) mb.calldefs( lambda calldef: calldef.call_policies is None ).exclude() This will work for your version of Py++ The way I am going to implement this is to check messages reported by > > "readme" method, to find all those I will define as problematic and exclude > > the declarations. > > > > P.S. I am a little bit busy these days, can you add new "feature > > request" to the project? Thanks. > > > Sure. > > I should try to produce a patch, I guess... :P > No, let me to handle this :-) PS: 0.9 was a _great_ release; it even produces code that compiles fine > with "off the shelf " Boost 1.33.1. That is a great thing; thank you so > much :-) > Thanks. -- Roman Yakovenko C++ Python language binding http://www.language-binding.net/ |
