Release calc version 2.11.0t5.2

This commit is contained in:
Landon Curt Noll
1999-10-05 21:12:27 -07:00
parent ea64a95b90
commit 8cabbd6fb4
8 changed files with 175 additions and 48 deletions

50
BUGS
View File

@@ -114,50 +114,14 @@ Known problems or mis-features:
**** 9 error(s) found \/++\/
9999: Ending regression tests
* On a Dec Alpha, make check for version 2.11.0t1 core dumps:
* On a Dec Alpha, using the Dec Alpha cc (not gcc) make check for
version 2.11.0t5.1 fails in the regression test:
From vandermj@molbio.sbphrd.com Fri Sep 24 06:15:28 1999
Subject: Re: calc version 2.11.0t1
Date: Fri, 24 Sep 1999 09:15:02 -0400
From vandermj@molbio.sbphrd.com Tue Oct 5 04:06:52 1999
Subject: Re: oops, try calc version 2.10.0t5.1
Date: Tue, 05 Oct 1999 07:06:30 -0400
It dies as follows:
Regular cc on Dec alpha, 'make check' dies with:
...
1802: a = surd(2,3)
1803: a == surd(2,3)
1804: surd_value(a) == 2+3i
Segmentation fault
dbx shows:
signal Segmentation fault at [objcall:155 +0xc,0x1200507b4] index = oap->actions[action];
(dbx) where
> 0 objcall(action = 14, v1 = 0x1400203d0, v2 = 0x1400203e0, v3 = (nil)) ["obj.c":155, 0x1200507b4]
1 comparevalue(v1 = 0x1400203d0, v2 = 0x1400203e0) ["value.c":2329, 0x12007f250]
2 o_eq() ["opcodes.c":2659, 0x12003728c]
3 calculate(fp = 0x14007fce0) ["opcodes.c":3614, 0x12003942c]
4 o_usercall(fp = 0x14007fce0, index = 39, argcount = 0) ["opcodes.c":2602, 0x120037080]
5 calculate(fp = 0x14003bfa0) ["opcodes.c":3622, 0x120039488]
6 evaluate(nestflag = 14) ["codegen.c":210, 0x120028768]
7 getcommands(toplevel = 536864560) ["codegen.c":158, 0x1200285bc]
8 getcommands(toplevel = 536864560) ["codegen.c":122, 0x1200284dc]
9 main() ["calc.c":279, 0x12001f61c]
Michel van der List reports that he was using the Dec compiler.
However On 25-Sep-99 9:18, Ernest Bowen wrote:
Last night I downloaded 2.11.0t1 and without changing Makefile
or anything else did a make and make check. Absolutely no
reported errors or warnings and the check "passed all tests".
The machine I'm using is a Dec 20 alpha running Digitial
UNIX V4.0B (Rev 564).
Ernest Bowen was using gcc.
We definitely want to hunt down this Dec's cc compiler issue. It
could very well be that gcc is letting something slide where as
Dec's cc compiler is compiling it faithfully into core dumping code.
The fault could very well be the calc code and not the compiler.
Any more information on calc core dumps compiled with Dec's cc compiler
would be most appreciated!
"": line 1706: Calling qfreenum with nozero links!!!