XBR2D96D@DDATHD21.BITNET (Knobi der Rechnerschrat) (06/29/90)
Hallo,
I've another F77 3.3 bug. As it seems very severe I post It to the list
and not to Calvin Vu (SGI) alone. If you compile/link/run the appended
program t1.f under 3.3 you will get the following two lines of output:
1 T
bla is false
As anybody can see, bla IS in fact TRUE, so the program lies. In the
release notes 3.3 FORTRAN 77 is stated:
(SCR 6822): Printing out LOGICAL*1 values results in 0/1 being printed
instead of F/T.
That is the cause for the '1 T' output line. Unfortunatelly there seems
to be an unknown problem with using LOGICAL*1 in IF statements. As this is
absolutely deadly for programs using LOGICAL*1 (e.g. for storage saving)
I feel this should be known to the "public" and fixed as soon as possible.
Regards
Martin Knoblauch
TH-Darmstadt
Physical Chemistry 1
Petersenstrasse 20
D-6100 Darmstadt, FRG
BITNET: <XBR2D96D@DDATHD21>
PS: Dear Calvin, sorry for posting this to everybody, but I feel it is
important news and I'm also not sure whether my mails make the way
to SGI directly.
----------------------------t1.f----------------------------------------------
program bug_6822
c
logical*1 bla
logical bla4
c
bla = .true.
bla4 = .true.
write(*,*) bla,bla4
c
if(.not.bla) write(*,*) 'bla is false'
if(.not.bla4) write(*,*) 'bla4 is false'
c
end
---------------------------end of t1.f-----------------------------------------calvin@dinkum.sgi.com (Calvin H. Vu) (07/03/90)
In article <9006291416.aa20103@VGR.BRL.MIL> XBR2D96D@DDATHD21.BITNET (Knobi der Rechnerschrat) writes: > program bug_6822 >c > logical*1 bla > logical bla4 >c > bla = .true. > bla4 = .true. > write(*,*) bla,bla4 >c > if(.not.bla) write(*,*) 'bla is false' > if(.not.bla4) write(*,*) 'bla4 is false' >c > end >---------------------------end of t1.f----------------------------------------- This is introduced in MIPS 2.0 release to allow logical*1 variables to contain integer*1 values (for VMS compatibility). There was an oversight in the implementation that caused integer operations on the logical*1 variables. This has been fixed in the 3.3 maintenance release (and in MIPS 2.10 release). Sorry about the inconvenience. - calvin ----------------------------------------------------------------------------- Calvin H. Vu | "We are each of us angels with only one Silicon Graphics Computer Systems | wing. And we can only fly embracing calvin@sgi.com (415) 962-3679 | each other."