Message-ID: <20130713145929.GA4321@cs.toronto.edu>
Date: 2013-07-13T14:59:29Z
From: Radford Neal
Subject: missing PROTECT() in src/main/arithmetic.c
In-Reply-To: <mailman.27.1373709607.15244.r-devel@r-project.org>
> From: Herv? Pag?s <hpages at fhcrc.org>
> at lines 651 & 653 (integer_binary function):
>
> if (code == DIVOP || code == POWOP)
> ans = allocVector(REALSXP, n);
> else
> ans = allocVector(INTSXP, n);
>
> There are calls to warningcall() later in the function, which can
> trigger garbbage collection.
>
> Looks like the typical scenario where it seemed pretty safe to not
> PROTECT in the original version of the function but became very
> unsafe 3 years later when the calls to warningcall() were added to
> the function.
Note that there is also a problem with a possible warning from myfmod,
which in turn is called from R_pow.
The call of myfmod from R_pow should probably be replaced by something
else, since as it is, the following undesirable behaviour occurs:
> (-Inf)^(1e16)
[1] Inf
Warning message:
probable complete loss of accuracy in modulus
I think issuing a warning for this is probably not a good idea, but if
a warning is issued, it certainly shouldn't be this one.
Radford Neal