Re: Automatic documentation generation

Patrick Decowski (decowski@mit.edu)
Sun, 21 Sep 1997 00:35:53 -0400 (EDT)


hi pasha and valery,

i think that this is exactly the problem. i noticed already a few
months ago that the arguments that are used when calling the strncmp()
functions in THtml are not checked against a NULL pointer. the
implementation of the strncmp() function seems to be different on the
different platforms. some of them check the arguments against NULL and
others don't. the GNU C lib doesn't check, so it affects your IRIX 6.2/gcc
and my Linux system. so for instance the htmlFile variable in
THtml::MakeClass() should be protected agains NULL.

regards,
patrick.

---------------------------------------------------------------------------
Patrick Decowski |
24-509 | Home: (617)625-9352
Massachusetts Institute of Technology | Office: (617)253-1779
77 Massachusetts Av | Fax: (617)253-1755
Cambridge, MA 02139-4307 |
---------------------------------------------------------------------------
http://web.mit.edu/decowski/www/home.html | e-mail: decowski@mit.edu
---------------------------------------------------------------------------

On Sun, 21 Sep 1997, Pasha Murat wrote:

>
> Hi Valery,
>
> I'm afraid that your guess might be true: the same (1_03) version
> of ROOT generates HTML on AIX 4.2 (ROOT compiled for AIX 4.1) just fine.
> So this effect depends on the platform (I'm experiencing a problem
> on IRIX 6.2 with gcc-compiled ROOT)... Hopefully Rene could try to
> reproduce it. I wonder if anybody had the same problem with the same
> ROOT "flavour":
>
> THtml html;
> gHtml->MakeClass("TBRIK")
>
> should work with all the defaults (works with ROOT/AIX 4.1/Xlc, but
> breaks with ROOT/IRIX 6.2/gcc).
>
> Thanks, Pasha.
> --------------------------------------------------------------------------------
> Valery Fine writes:
> > On 20 Sep 97 at 16:19, Pasha Murat wrote:
> >
....
> >
> > I saw such sort of the problem if the local implementation
> > of the Run-time str() family of the C built-in functions doesn't
> > accept the ZERO pointers as the input parameters and the user code
> > doesn't check it in advance as well. May be we missed something.
> >
> > Valery
> > Dr. Valeri Faine (Valery Fine)
> > ------------ ------------- Phone: +41 22 767 4921
> > CERN FAX : +41 22 767 7155
> > CH-1211 Geneva, 23 mailto:fine@mail.cern.ch
> > Switzerland http://nicewww.cern.ch/~fine
> >
>