Hi! I looks (to me) like ape/math.h for 386 is missing the following two lines: #define isnan(x) isNaN(x) #define isinf(x) isInf(x, 0) They are present in /{68020|arm|arm64|power|sparc|sparc64|amd64}/include/ape/math.h. It even looks like the lines are present for 386 in the stock Bell Lab's distro. Any known reason for this irregularity? isnan() is used by Netsurf's SVG-library. /Jonas A --- orig_math.h Sat Oct 5 02:33:06 2019 +++ math.h Fri Jan 1 01:30:47 2021 @@ -73,4 +73,7 @@ } #endif +#define isnan(x) isNaN(x) +#define isinf(x) isInf(x, 0) + #endif /* __MATH */

```
On Fri, 1 Jan 2021 23:34:27 +0200, Jonas Amoson <jonas.amoson@gmail.com> wrote:
> Hi!
> I looks (to me) like ape/math.h for 386 is missing the following two lines:
>
> #define isnan(x) isNaN(x)
> #define isinf(x) isInf(x, 0)
>
> They are present in
> /{68020|arm|arm64|power|sparc|sparc64|amd64}/include/ape/math.h. It
> even looks like the lines are present for 386 in the stock Bell Lab's
> distro. Any known reason for this irregularity? isnan() is used by
> Netsurf's SVG-library.
>
> /Jonas A
While you're touching it, do you think you can unify the math.h stuff,
so they don't go out of sync?
/$objtype/include/ape/_math.h:
arch-specific math.h bits
/sys/include/ape/math.h:
arch-independent math.h, including <_math.h>
```

```
Quoth Ori Bernstein <ori@eigenstate.org>:
>
> While you're touching it, do you think you can unify the math.h stuff,
> so they don't go out of sync?
>
> /$objtype/include/ape/_math.h:
> arch-specific math.h bits
> /sys/include/ape/math.h:
> arch-independent math.h, including <_math.h>
>
...never mind, I looked and did it. the only differences
between the copies were bugs.
```