Date: Sat, 11 Sep 1999 19:04:21 -0700 (PDT) From: lou@man-97-187.reshall.berkeley.edu To: FreeBSD-gnats-submit@freebsd.org Subject: bin/13703: libf2c Message-ID: <199909120204.TAA10289@man-97-187.reshall.berkeley.edu>
index | next in thread | raw e-mail
>Number: 13703
>Category: bin
>Synopsis: MCNP compilation problem
>Confidential: no
>Severity: serious
>Priority: medium
>Responsible: freebsd-bugs
>State: open
>Quarter:
>Keywords:
>Date-Required:
>Class: change-request
>Submitter-Id: current-users
>Arrival-Date: Sat Sep 11 19:10:00 PDT 1999
>Closed-Date:
>Last-Modified:
>Originator: Tak Pui Lou
>Release: FreeBSD 3.2-RELEASE i386
>Organization:
UCBNE
>Environment:
>Description:
When I compiled MCNP, I can't get one of the subroutine working probably. This subroutine tries to read the input and write it to another file. When the input has a line begins with "like", it backspace and read the previous line again and write the previous line with some changes to the output file. But, it seems to be confuse about the file formats after that. The input is ascii. I don't know about the output but it seems to be binary but become ascii after "like" card or vice-versa. (I am not sure because it is a hugh program.) This happens to all FBSD versions I have used. There is no error message when you compile it, but the program won't run with "like" card.
>How-To-Repeat:
To show you what is wrong, I attach a small portion of the subroutine likebt.f.
c
if(mm.eq.2)go to 150
c
c when "like" is seen, remove the card from iu1.
if(nwc.eq.3)return
if(nwc.ne.1)go to 10
likef=1
mxa=mxa-1
backspace iu1
call nxtsym(klin,' ',1,it,iu,1)
if(klin(it:iu).eq.'like')backspace iu1
endfile iu1
return
c
c get the material and density and copy the geometry of cell "m".
10 if(nwc.ne.2)go to 120
do 20 nc=1,mxa
20 if(mod(ncl(lncl+nc),100000).eq.iitm)go to 30
go to 110
30 ncl(lncl+mxa+1)=ncl(lncl+mxa+1)+100000*(ncl(lncl+nc)/100000)
rewind iu2
rewind iu1
do 50 i=1,nc
40 read(iu1,'(a5,a80)')ht1,hl1
50 if(ht1(1:1).eq.'&')go to 40
ipp=4
nw=0
hl2=' '
call nxtsym(hl1,' ',1,it,iu,1)
go to 80
60 read(iu1,'(a5,a80)',end=110)ht1,hl1
if(ht1(1:1).ne.'&')go to 100
iu=0
80 call nxtsym(hl1,' ():#$&',iu+1,itl,iu,1)
if(iu.eq.0)go to 60
if(hl1(itl:itl).eq.'$'.or.hl1(itl:itl).eq.'&')go to 60
nw=nw+1
hh=hl1(itl:iu)
if(nw.eq.1)read(hh,'(bn,i30)')mats
if(nw.eq.2.and.mats.ne.0)read(hh,'(bn,e30.0)')rhos
if(nw.eq.1.or.nw.eq.2.and.mats.ne.0)go to 80
k=index('():#',hh(1:1))
ki=kdata(hh)
if(k.eq.0.and.ki.eq.0)return
if(k.eq.4)ncomp=ncomp+1
if(ipp+2+iu-itl.le.80)go to 90
write(iu2,'(a80)')hl2
ipp=4
hl2=' '
90 hl2(ipp+2:ipp+2+iu-itl)=hh
ipp=ipp+2+iu-itl
go to 80
100 itl=-1
return
110 itl=-2
return
c
c copy the exceptions onto iu2.
120 if(hitm.eq.'=')return
if(likef.ne.1)go to 140
if(hitm.eq.'mat')likef=2
if(hitm.eq.'rho')likef=3
if(likef.ne.1)return
n=0
if(hitm.ne.':'.and.hl2(ipp:ipp).ne.':')n=1
if(ipp+3+nitm.le.60.or.n.eq.0)go to 130
write(iu2,'(a80)')hl2
ipp=4
hl2=' '
130 hl2(ipp+n+1:ipp+n+nitm)=hitm
ipp=ipp+n+nitm
return
c
c save mat or rho values.
140 if(likef.eq.2)mats=iitm
if(likef.eq.3)rhos=ritm
likef=1
return
c
c when the "like" card is all read, finish the expanded card.
150 if(nwc.lt.2)go to 210
backspace iui
c
c copy any cell parameters of cell "m".
if(itl.eq.-2)go to 200
if(itl.eq.-1)go to 190
iu=itl-1
go to 170
160 read(iu1,'(a5,a80)',end=200)ht1,hl1
if(ht1(1:1).ne.'&')go to 190
iu=0
170 call nxtsym(hl1,' ()#$&',iu+1,itl,iu,1)
if(iu.eq.0)go to 160
if(hl1(itl:itl).eq.'$')go to 160
if(ipp+4+iu-itl.le.80)go to 180
write(iu2,'(a80)')hl2
ipp=4
hl2=' '
180 hl2(ipp+2:ipp+2+iu-itl)=hl1(itl:iu)
ipp=ipp+2+iu-itl
go to 170
c
c skip over the rest of iu1.
190 read(iu1,'(a5,a80)',end=200)ht1,hl1
go to 190
200 backspace iu1
c
c get ready to read the expanded card from iu2.
write(iu2,'(a80)')hl2
endfile iu2
rewind iu2
jui=iu2
write(hlin,'(i5,i6)')icn,mats
if(mats.ne.0)write(hlin(13:33),'(e21.14)')rhos
210 likef=0
return
end
>Fix:
Recompile libI77 under /usr/src/lib/libI77 with the Makefile in /usr/src/libf2c with the -DNON_ANSI_RW_MODES removed from the CFLAG. I believe the problem is wrong configuration of libI77. I would also like to ask you if you can put -DWANT_LEAD_0 in the CFLAG. The reason is a lot of FORTRAN compilers (e.g. from SUN, DEC, SGI, ...) have a leading zero for decimal number smaller than 1. I am pretty sure MCNP is working probably because it can be compiled under many machines.
>Release-Note:
>Audit-Trail:
>Unformatted:
To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-bugs" in the body of the message
help
Want to link to this message? Use this
URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199909120204.TAA10289>