pop@linus.mitre.org (Paul Perry) (11/01/90)
A problem, a fix, and a question. When building isisv2.1 for a VAX3100 Ultrix-32 V3.0 (Rev 64) I ran into this error: cd util; make MCHTYPE=VAX MCHDEPCFLAGS="-DBYPASS" OPLIBS= CC=cc OPTIM="-O" ; cd .. cc -DBYPASS -DVAX -I../../include -c -O ../../util/long_haul.c "../../util/long_haul.c", line 2663: warning: ambiguous assignment: assignment o p taken "../../util/long_haul.c", line 2663: warning: old-fashioned initialization: use = "../../util/long_haul.c", line 2663: syntax error "../../util/long_haul.c", line 2663: warning: old-fashioned initialization: use = "../../util/long_haul.c", line 2663: warning: illegal combination of pointer and integer, op = *** Error code 1 which is easily fixed by adding a space after the = sign on that line. ISIS then compiles and work fine. The question is: if I build one isis on a sun4 with the BYPASS and a second isis on a VAX3100 withought the BYPASS flag are they going to communicate well ? I ask because I noticed that the grid demo didn't seem to behave as expected when I had this setup. I realize I am being vague here. I have since rebuilt isis with the BYPASS flag on and all is ok. -- Paul O. Perry MITRE Corporation Phone: (617) 271-5230 Burlington Road ARPA: pop@mitre.org Bedford, MA 01730 UUCP: ...{decvax,philabs,genrad}!linus!pop
rcbc@cs.cornell.edu (Robert Cooper) (11/01/90)
In article <125054@linus.mitre.org>, pop@linus (Paul Perry) writes: >When building isisv2.1 for a VAX3100 Ultrix-32 V3.0 (Rev 64) I ran >into this error: >"../../util/long_haul.c", line 2663: warning: old-fashioned initialization: use >= >which is easily fixed by adding a space after the = sign on that line. >ISIS then compiles and work fine. Thanks, we'd already caught that one, but assumed most compilers would only give a warning not an error so we didn't post the fix. Here's the fix: 2663c2663 < int i, j=*addrlen; --- > int i, j = *addrlen; Since most of us learnt C after they replaced the dangerous i =+ 1 syntax with the more sensible i += 1, we often forget to leave the requisite space. >The question is: if I build one isis on a sun4 with the BYPASS and a >second isis on a VAX3100 withought the BYPASS flag are they going to >communicate well ? I ask because I noticed that the grid demo didn't >seem to behave as expected when I had this setup. I realize I am >being vague here. I have since rebuilt isis with the BYPASS flag on >and all is ok. You must compile all binaries with the same setting of the BYPASS flag. We might put in a run-time sanity check for this, but at some point BYPASS will be standard and non-BYPASS will be removed as an option anyway. -- Robert Cooper