On Sun, 20 Sep 2015, Allin Cottrell wrote:
On Sun, 20 Sep 2015, oleg_komashko(a)ukr.net wrote:
> Dear Allin,
> the same script goes ok on xp with p=1 and q=1;
> also I copied the armax() code to gretl script file
> with the same result
Sorry, but by "the same result" do you mean that running the revised version
of the armax() function that I sent to the list, there's still a crash on XP
when p = q = 2? (Though not with p = q = 1?)
> Of course xp is a very old stuff but some students in post-
> soviet countries and China still use it
> I only may guess it's something with arithmetic in dll
In my experience an actual crash always indicates a bug in gretl, even if
it's not triggered on all OSes. (Tempting as it might be to believe that it's
Microsoft's fault!) Well, one more thought: if you run gretl.exe with the
--debug flag, do you see anything that might be of interest on stderr? (That
is, in the accompanying console window.)
I found something in our lexer that, under certain very specific
conditions, might produce a crash (and quite likely in an
OS-dependent way).
The fix is in CVS and new snapshots. I wonder if this might deal
with the problem Oleh found on Windows XP?
Allin