Hello All,

A fellow named Ingo van Lil has been helping me recently with finding
and fixing bugs in Wcalc. It occured to me that our conversation
probably ought to be on the list, just to help make sure any design or
feature decisions work for everyone involved, and to solicit help and
comment on the problems.

On Sat, Aug 30, 2003 at 03:42:37AM +0200, Ingo van Lil quoth:
> It wasn't meant to replace your current parser, it was just an example
> how to assign different precendece to one operator depending on the
> context.

Yeah, I know... this is a problem that I think will require much more
fiddling.

> I'm currently trying to create a simpler parser that can do implicit
> multiplication, but I'm not really successful. It compiles without
> reporting any conflicts and it shows the correct result in most cases,
> but it fails (parse error) if I use more than one factorial operator
> (e.g. 5!!). It probably tries to parse the exclamation marks as 'NOT'
> operators and then finally wonders at not finding any operand to apply.

Good luck! It took me lots of rewrites and some hard thinking to get the
parser as far as I've gotten it... but if you get something a little
simpler or a little less ambiguous, I'm all ears.

> > > Strange discussion. I can't really follow Jerry's way of thought in his
> > > sentence "IMHO, if you type 5* and then another * it should calculate
> > > 25".
> >
> > I didn't follow his logic either. It might have been a GUI-only thought
> > of his... I'm not sure. I may try getting rid of it, and recoding it to
> > do what he thinks it ought to in the GUI. What do you think?
>
> I don't know how the GUI works, but I could imagine hitting the '*' key
> n-times being interpreted as "x^n". In the same way multiple '/' strokes
> could be interpreted as n-th root. Sounds like a nice feature.

Yes indeed! I'll see what I can do about that one... getting the buttons
to do fun things beside simply insert characters into the expression is
a little involved, but I'll work on it.

> > Oh, another thought that occurred to me... should we continue this
> > conversation on the Wcalc mailing list, just to encourage people to
> > comment?
>
> Sure, why not? Just reply to the list next time (and maybe put some
> explaining words for the other subscribers in front of it).

Done.

> By the way, I found yet another small (and probably easy to fix) bug:
>
> -> foo=2
> foo = 2.000000
> -> 4foo
> = 42
>
> and worse:
>
> -> bar=-4
> bar = -4.000000
> -> 2 bar
> = -2

Ahhh... I knew flatten() was going to bite me someday. Okay, I fixed it
in CVS. Basically, instead of inserting the value/string of the variable
into the expression, it now inserts it inside of parenthesis. End
result: should work as expected now.

~Kyle
--
Anyway, have fun. And don't bother reporting any bugs for the next few days. I
won't care anyway.
-- Linus Torvalds, when kernel 2.4 came out