> that allows it to be built both as a single binary *or* with one class per binary.
 
ah, of course that makes sense.
 
> with multi-binary style I can do [declare -lib foo/bar] whereas with the single-binary style that won't work.
 
I'm not sure I understand. Can give an example?
 
Gesendet: Sonntag, 12. Mai 2019 um 16:51 Uhr
Von: "Jonathan Wilkes" <jancsika@yahoo.com>
An: "Christof Ressi" <christof.ressi@gmx.at>
Cc: Pd-List <pd-list@lists.iem.at>
Betreff: Re: Aw: Re: Re: Re: [PD] Full stop '.' in class namespace
 
> On Sunday, May 12, 2019, 12:27:33 AM PDT, Christof Ressi <christof.ressi@gmx.at> wrote:
 
>> So why aren't all (or at least nearly all) external developers simply using "setup" as the entry point?

> yes, I've asked that myself and I think they just could. Have a look at the Max part of pd~: the setup function is simply called 'main'.
 
Here's the probable reason-- current practice is aimed at the maximally complicated library which has a makefile that allows it 
to be built both as a single binary *or* with one class per binary. In that case you must have uniquely-named setup functions 
per class that can be called from a central library setup.
 
Which I suppose would make more sense *if* the developer's choice of single- or multi-binary style was merely an 
implementation detail. But even with your patches it isn't-- with multi-binary style I can do [declare -lib foo/bar] whereas with 
the single-binary style that won't work.
 
-Jonathan