I'm getting #<a FLOATING-POINT-UNDERFLOW> sometimes when communicating
with cffi calls. It's farely inconsistent - so I'm trying to debug it.
Any suggestions? (the same code works fine in sbcl)
incidentally, outside of several systems being unusable thanks to a
shortage of 64 bit integers, hal and dbus hooks are almost entirely working.
(I'm also not sure how to handle null-terminated string arrays)
No idea if anyone is interested - I've just been porting the libhal
demos (slowly) over to ecl as a bit of an exercise. If anyone is -
any suggestion where I could post it? It's not really that small.
Also I could probably convert it to uffi (or ecl ffi) if need be -
automatically-built swig code was not usable without heavy modification
so I don't mind.
From: Juan Jose Garcia-Ripoll <juanjose.garciaripoll@go...> - 2008-09-01 08:33:23
On Mon, Sep 1, 2008 at 6:28 AM, Teunis Peters <teunis@...> wrote:
> I'm getting #<a FLOATING-POINT-UNDERFLOW> sometimes when communicating
> with cffi calls. It's farely inconsistent - so I'm trying to debug it.
It is probably related to the library you are interfacing with, which
does not care much about floating point precision. You can use
(SI:TRAP-FPE 'FLOATING-POINT-UNDERFLOW NIL) to ignore those events
Instituto de Física Fundamental
CSIC, Serrano, 113, Madrid 28040 (Spain)