Re: [bug-gawk] useless array length abstraction

[Andrew J. Schorr]

On Sun, Jan 06, 2019 at 05:30:16AM -0700, address@hidden wrote:
> Hi.
> 
> "Andrew J. Schorr" <address@hidden> wrote:
> 
> > I propose the attached patch that removes the alength function and cleans
> > up this issue.
> 
> This is OK, but is not quite complete. There are, I think, two uses
> in symbol.c and one in gawkapi.c where direct access to xxx->table_size
> should be replaced by assoc_empty.  Please fix those as well and then
> push to master.

Will do. Is it OK if the patch to reset the array type to null when the
last element is deleted is consolidated in the same patch, or should that
be committed separately? And does that one belong in master as well,
or might you consider putting it in stable?

> > But taking another step back, I'm not sure why these array methods are
> > implemented as an array of function pointers instead of as a struct
> > with a field for each method.
> 
> This would be a step forward.  If you can work up a patch for that for
> review, I'd appreciate it.

I will see how difficult this is.

Regards,
Andy