Re: Re: Re: Windows/MSVC: install_builtins, octave_interpreter_ready not found

[Christoph.Mayer]

> Von: John W. Eaton [mailto:address@hidden
> Gesendet: Mittwoch, 28. November 2007 17:43
> An: Mayer, Christoph
> Cc: address@hidden
> Betreff: Re: Re: Windows/MSVC:
install_builtins,octave_interpreter_ready
> not found
> 
> On 28-Nov-2007, address@hidden wrote:
> 
> | I propose the following patch (see below):
> |
> |     2. adding more OCTINTERP_API macros to toplev.h
> |
> | At the moment I need only do_octave_atexit(), but for more
flexibility
> I'd like to have access all global variables and functions in toplev.h
> 
> Why do you need access to these symbols?  We could just export all the
> symbols (I think that is what happens with most Unixy compilers) but
> then you do realize that just because the symbols are exported, they
> are still not part of any "official documented API" and are subject to
> change or removal at any time?
> 

I need these symbols because I have an application which uses more than
one octave interpreter; I have implemented context switching, where each
octave interpreter has its own symtabs. If there is another more
standard way to have multiple octave interpreters in one application I
would like to use it.

In order to hide these "not officially documented" symbols one could add
an "OctaveInterpreter" class, which performs context switches, lets
users define and retrieve symbols, etc. to octave. 

> | I have now succeeded in compiling parts of octave-2.9.13
(octave.exe,
> liboctave, octinterp, libcruft) under MSVC2005 after manually adding
>       #define WINVER 0x500
> |     #define _WIN32_WINNT 0x500
> | to libcruft/misc/cquit.c
> 
> Did you mean 2.9.17?  If not, why are you using 2.9.13?
> 
I mean 2.9.13. First I managed to partially compile 2.9.13, then 2.9.17.

Christoph