Jan,

It's great that the Solaris port of Coda is progressing, I sucessfully
compiled and installed the kernel module for Solaris.  Your suggestion of
-ly did not work, however, which is strange, since:

asr > nm /usr/lib/liby.so | grep yyerror
[59]    |      1724|      48|FUNC |GLOB |0    |9      |yyerror

Within the asr directory, here is what my make looks like:

asr > make clean ; make
rm -f *.o *.a  *.d
rm -f *.client.cc *.client.c *.server.cc *.server.c *.multi.cc *.multi.c
*.print.c 
rm -f parser  resolver_lexer.c resolver_parser.c y.tab.h resolver_parser.h
core
rm -f chdrinstall rp2hdrinstall libinstall bininstall scriptinstall \
      c_bin_install c_sbin_install s_bin_install s_sbin_install 
rm -f *.l.c *.pt.c *.l.o *.pt.o
yacc -d resolver_parser.y && mv y.tab.c resolver_parser.c && mv y.tab.h
resolver_parser.h
yacc -d resolver_parser.y 
mv -f y.tab.c resolver_parser.c
c++ -fno-exceptions -x c++ -g -O2   -MD -DHAVE_CONFIG_H
-I/tmp/codawork/coda -I. -I/tmp/codawork/coda/include -c resolver_parser.c
flex  -t resolver_lexer.l > resolver_lexer.c
gcc -fno-exceptions -g -O2   -MD -DHAVE_CONFIG_H  -I/tmp/codawork/coda -I.
-I/tmp/codawork/coda/include   -c -o resolver_lexer.o resolver_lexer.c
resolver_lexer.l:55: warning: prototype for `yywrap' follows
resolver_lexer.l:34: warning: non-prototype definition here
c++ -fno-exceptions -g -O2 -MD -DHAVE_CONFIG_H  -I/tmp/codawork/coda -I.
-I/tmp/codawork/coda/include   -c -o ruletypes.o ruletypes.cc
c++ -fno-exceptions -g -O2 -MD -DHAVE_CONFIG_H  -I/tmp/codawork/coda -I.
-I/tmp/codawork/coda/include   -c -o resolver.o resolver.cc
gcc -fno-exceptions -g -O2   -MD -DHAVE_CONFIG_H  -I/tmp/codawork/coda -I.
-I/tmp/codawork/coda/include   -c -o wildmat.o wildmat.c
gcc -fno-exceptions -g -O2   -MD -DHAVE_CONFIG_H  -I/tmp/codawork/coda -I.
-I/tmp/codawork/coda/include   -c -o path.o path.c
c++ -fno-exceptions -R/usr/local/lib resolver_parser.o resolver_lexer.o
ruletypes.o resolver.o wildmat.o path.o ../kerndep/libkerndep.a
../util/libutil.a /tmp/codawork/coda/lib-src/base/libbase.a -lfl -lnsl
-lsocket -lc    -L/usr/lib -ly -o parser
Undefined                       first referenced
 symbol                             in file
yyerror(char const *)               resolver_parser.o
ld: fatal: Symbol referencing errors. No output written to parser
collect2: ld returned 1 exit status
make: *** [parser] Error 1
rm resolver_lexer.c

As for libdb, is the use of libdb 1.85 a final decision made by the Coda
project?  I am having trouble finding a port of libdb for Solaris 7, I can
not find any packages, and the only source I am able to locate is for
Solaris 2.2.  www.sleepycat.com seems to be the current distribution
center for libdb as well as other old software from UCB, and this is the
site that has a port to solaris 2.2, which does not compile on Solaris 7.

mohan


On Fri, 16 Jun 2000, Jan Harkes wrote:

> On Fri, Jun 16, 2000 at 12:10:15PM -0400, Mohan Khurana wrote:
> > codalist,
> > 
> > Does anyone have any information on how the solaris port of coda is
> > progressing?  On the configure script that I execute, it tells me that it
> > doesn't want to use a current version of libdb, claiming that certain
> > functionality will not be available for coda.  I'm really new to coda
> > and I am still trying to figure out the general way it works, but I am
> > wondering if efforts are underway to get coda to use NDBM rather than
> > UCB's db.  I am using the source tree obtained from coda's CVS repository.
> 
> Some of the configuration databases (users/groups) are simply copied to
> all servers, we try to use the same database format on all servers. So
> we picked db1.85, because it is available on most (all?) platforms, is
> heterogeneous, etc.
> 
> If you use ndbm, you will probably only be able to use servers on
> solaris. The checks in configs/coda_macros.m4 tend to favor db1.85, so
> that's probably the reason it won't pick up ndbm that easily.
> 
> > I am interested in using coda as an NFS replacement for one of my
> > projects.  The network where coda is to be used it a heterogeneous one
> > however, including FreeBSD, Linux, OpenBSD, and a lot of Solaris.  I was
> > hoping someone could shed some light as to how the Solaris port is
> > progressing.  There doesn't seem to be any mention of Solaris in the
> > HOWTO.
> 
> The userspace code for the client and server is as solid as on any other
> UNIX platform. The kernel code for the client still has no attribute
> caching so it might be a bit slower, but is already very reliable.
> 
> > Here is the error that I get when doing make client-install on Solaris:
> > 
> > c++ -fno-exceptions -R/usr/local/lib resolver_parser.o resolver_lexer.o
> > ruletypes.o resolver.o wildmat.o path.o ../kerndep/libkerndep.a
> > ../util/libutil.a /tmp/codawork/coda/lib-src/base/libbase.a -lfl -lnsl
> > -lsocket -lc    -o parser
> > Undefined                       first referenced
> >  symbol                             in file
> > yyerror(char const *)               resolver_parser.o
> > ld: fatal: Symbol referencing errors. No output written to parser
> 
> Does:
> 
> ../util/libutil.a /tmp/codawork/coda/lib-src/base/libbase.a -lfl -lnsl \
>     -lsocket -lc -ly   -o parser
>                  ^^^
> make it link correctly?
> 
> Jan
> 
>