Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!utgpu!water!watmath!clyde!cbosgd!ncr-sd!randall
From: randall@ncr-sd.UUCP
Newsgroups: comp.bugs.misc
Subject: Bug in dc?
Message-ID: <1800@ncr-sd.SanDiego.NCR.COM>
Date: Fri, 2-Oct-87 18:34:59 EDT
Article-I.D.: ncr-sd.1800
Posted: Fri Oct  2 18:34:59 1987
Date-Received: Sat, 3-Oct-87 11:47:04 EDT
Reply-To: randall@ncr-sd.SanDiego.NCR.COM (0000-Randall Rathbun)
Organization: NCR Corporation, Rancho Bernardo
Lines: 34
Keywords: dc, segmentation error

Running the short dc routine below to find the gcd of two numbers
appears to work for small numbers. The command line is:

%dc gcd

r & s are passing parameters, returned in lowest terms
initiate by: l7x
======================================================
[lx_1*sx]s1[ly_1*sy]s2[lyszlxsylzsx]s3[1sz]s4
[lxly%szlysxlzsy1lz>5lz0=6lz1=4]s5[lxsz]s6
[lrsxlssylx0>1ly0>2lxly>3l5xlrlz/srlslz/ss]s7
?sr?ssl7x
[r,s in lowest terms:]Pc
[r= ]Plrpc
[s= ]Plspc
q
======================================================

When r & s are about 100 digits or so in length, an error message appears:

%Segmentation fault (core dumped)

What is wrong? Is the stack being recursively called until it overflows?
I see nothing wrong in the subroutine above. Can anyone find out what is
the problem and let me know? (e-mail is preferable) Is this a bug in dc?

I am working with rational points on cubic surfaces, and need to have this
problem fixed, to investigate the action of the chord-tangent process on
creating new E(Q) pairs of the infinite birational pairs group.

reply to: Randall.Rathbun@thor.SanDiego.NCR.COM
  or      randall@thor.sandiego.NCR.COM

thanks sincerely   -   Randall