Re: saveing/loading symbol table of annymous functions

[David Bateman]

John W. Eaton wrote:
> On 14-May-2007, David Bateman wrote:
> 
> | In fact the function handle workspace doesn't follow the function
> | handles class 17 block. Rather the header has an offset into the file
> | where the sub-system workspace is located and all function handle
> | workspaces have a field MCOS that is a signature and an offset into the
> | sub-system workspace for the workspace of each of the individual
> | function handle workspace. There are still some things I don't
> | understand in the file format for function handles, but I think I have
> | the loading of function handles working correctly for matlab files. I
> | think I'd like to see this included in octave and get a little bit of
> | feedback on issues people have to try and better understand the format
> | to allow saving to work.
> | 
> | The attached patch also includes a load/save for the octave file formats
> | for function handles that respects the path to system files, saves the
> | anonymous handles workspace, includes load/save in mat-file format for
> | inline functions and updates the Ffunctions function to print the
> | workspace and be a bit more compatible.
> 
> Do you think this change is safe for 2.9.11?  It seems it should be OK
> since it just adds new functionality, but I can't easily tell whether
> the patch touches much other existing code in a way that could be
> risky.  If you agree that it is relatively safe, please check it in.

I think it should be ok, since as you say it just adds functionality, or
it fixes functionality that wasn't working correctly for octave's own
function handle load/save file formats.

> 
> | + std::string
> | + octave_system_path (void)
> | + {
> | +   return Vsystem_path;
> | + }
> 
> Maybe this should be a static function in the load_path class?

If you prefer,,

D,


> 
> Thanks,
> 
> jwe
>