[prev in list] [next in list] [prev in thread] [next in thread] 

List:       gnuplot-info-beta
Subject:    Re: isnan()
From:       Petr Mikulik <mikulik () monoceros ! physics ! muni ! cz>
Date:       2002-08-29 10:20:10
[Download RAW message or body]

> snan(double val) is in AES XPG3 XPG4 BSD and IEEE standards
> There are systems that don't match any of those,

Neither gcc on some systems conform to these standards.

> No. But it should.  On most, maybe all platforms, ISNAN should be
> #defined to isnan(), _isnan() or whatever it's called there.  Only if
> nothing else is found we should use this fallback definition.

It would be nice if you can add this feature to configure. We can change
config/makefile.* afterwards.

> How is NaN treated in text data files?

I think that if a data point has any coordinate NaN, then it should be
coordinate->type = UNDEFINED. It is not; it would be worth fixing it.

> > Further it seems that autoscaling goest to an infinite look if there is Inf
> > (in a binary datafile). Any idea why this happens?

> Because the infinity is not detected properly, I'd say.  May happen if
> VERYLARGE is not defined to infinity, even though the platform does
> support computations with infinities.

Could you please try to fix this?


> Handling of NaN and Infinities is one area where the C programming
> language really need some of the polishing it got in C99.

Hm, it seems you both speak about these standards... can you please give me
some pointers which you use for reference? (I could google it, but that
gives too much answers.)

---
Petr Mikulik


[[[[ unsubscribe from info-gnuplot-beta via majordomo@dartmouth.edu ]]]]

[prev in list] [next in list] [prev in thread] [next in thread]