Re: [Gm2] Numerical library in Modula-2 - IO

[Gaius Mulley]

Michael Riedl <address@hidden> writes:

> Gaius,
>
> I need some small support.

Hi Michael,

I think it might be a good idea for gm2 to expand its knowledge of
builtins to include all the floating point number classification
functions

http://www.gnu.org/software/libc/manual/html_node/Floating-Point-Classes.html

this way the compiler will generate tight code by expanding the builtins
internally.  I'll work on a fix for trunc and back port

regards,
Gaius

> In order to prepare the publication of the Numerical library I
> reworked some parts of my IO library
> so that it is sitting "on top" of the ISO lO libraries - this to
> ensure that someone using the library
> can do so whilst using ISO IO routines elsewhere in his project. In
> general the routines do little IO - mostly error massages have to be
> taken care about - but for the output of matrices the picture is
> clearly different, the same for some optimizers when used in long
> running applications.
>
> I achieved that in having two versions of Streams.mod and
> FileSystem.mod and a small amount of files now having some
> preprocessor directives so that they can be used either "standalone"
> or "on top of ISO IO".
>
> What I would need now is a library to check for NAN and INF for the 12
> byte REAL in GM2. Attached my internal version for 4 Byte and 8 Byte
> Reals.
>
> As soon as this is fixed I think both libraries are in a state to put
> a first draft out to the community.
>
> Thanks in advance
>
> Michael