Re: Parsing timestamps?

On Mon, 23 Jun 2025 13:34:08 +0000, mhx wrote:

On Mon, 23 Jun 2025 10:02:44 +0000, minforth wrote:
>

On Mon, 23 Jun 2025 5:40:37 +0000, mhx wrote:
>

On Sun, 22 Jun 2025 21:27:40 +0000, minforth wrote:
>
[..]
Do you mean your compiler automatically handles/allows combinations
like
... 22e-12 69. A{{ ( F: -- a ) ( D: -- b ) ( M: -- c ) SPIN ...
>
I found that handling mixed types explodes the code that needs
to be written for a simple compiler like, e.g., Tiny-KISS . It
would be great if that can be automated.

I don't know if I got you right, because as previously
defined, SPIN expects three integers on the data stack.

>
I was indeed too hasty. If items are stacked, no type conversion
is needed if they are only reordered. (Reordering needs
no code anyway, as it is a only a memo to the compiler.) The
problems only arise when an cell must be translated to a complex
extended float, or when using floats to initialize an arbitrary
precision matrix.

This is too vague for me, but perhaps it's not important anyway.
In any case, my compiler also supports Z: type complex locals.

Do you really support matrix and string type locals? The former
I only do for arbitrary precision, the latter can be handled
with  DLOCALS| .

I use a matrix stack (a depth of 10-20 items is usually sufficient).
By default, matrix elements are 32-bit sfloats. Matrix locals
(and mvalues) are just small structs containing matrix information
and a pointer to the heap memory allocated for all matrix elements.
Dynamic strings are simply 1-dimensional character matrices.
--