Re: declarations for built-in DEFUN functions

[Daniel J Sebald]

On 09/18/2012 03:23 PM, John W. Eaton wrote:
> On 18-Sep-2012, Rik wrote:
>
> |>
> |>  Message: 7
> |>  Date: Tue, 18 Sep 2012 15:41:47 -0400
> |>  From: "John W. Eaton"<address@hidden>
> |>  To: octave maintainers mailing list<address@hidden>
> |>  Subject: declarations for built-in DEFUN functions
> |>  Message-ID:<address@hidden>
> |>  Content-Type: text/plain; charset="us-ascii"
> |>
> |>  Some questions:
> |>
> |>  * Nearly all source files already have corresponding header files.
> |>  Should we include the -defun-decls.h files there instead of in the
> |>  .cc files?
> |
> | I would expect to find exported function prototypes in header files so this
> | makes sense to me.
>
> What about for the functions in the libinterp/corefcn directory?
> Those don't currently have corresponding header files.  Should they,

I'd say no.  Reason being that the contents of these files really isn't 
OOP code using the C++ constructs.  It is the class definitions of C++ 
that typically have a header-file/code-file combo.  I think they could 
be called .c rather than .cc for valid reasons.  Similarly, the lex.cc 
code could be compiled under C which would eliminate some cast warnings.


> or should we just use the -defun-decls.h files for them?  There are 66
> files there, and more than half of those files provide only a single
> function.  So if we create header files for each of them we would be
> adding another 66 header files in addition to the 66 generated
> -defun-decls.h files.  Or, I suppose we could treat those files
> specially and generate foo.h instead of foo-defun-decls.h.
>
> Another option I considered was to just lump all these declarations
> together in builtins.h and not generate separate header files for each
> source file that defines a built-in function.

Yes that sounds very good.

Dan