On Mon, 23 Aug 2004, Mark Taylor wrote:
> Date: Mon, 23 Aug 2004 15:10:13 +0100
> From: Mark Taylor <[log in to unmask]>
> Reply-To: Starlink development <[log in to unmask]>
> To: [log in to unmask]
> Subject: Re: kappa problems (was Re: datacube)
>
> On Mon, 23 Aug 2004, David Berry wrote:
>
> > > > Okay, I don't know why I didn't trip over this before during 'make world'
> > > > but it's not that ndfpack_mon isn't built, it's that it doesn't build.
> > > > ...
> > > > ld: multiple definitions of symbol _matherr
> > > > /star/lib/libtcl8.2.a(tclMtherr.o) definition of _matherr in section (__TEXT,__text)
> > > > /star/lib/libprm_a.a(num_han.o) definition of _matherr in section (__TEXT,__text)
> > > > klunk{star}:
> > > >
> > > > Anyone else come across this one?
> > >
> > > Yes, I see the same problem under Cygwin. Only solution I came up with was
> > > to "#ifdef __CYGWIN32__" out the Tcl version.
> >
> > Wasn't this (the matherr problem) fixed by a mod to tclsys which Mark
> > submitted last week?
>
> there certainly was a matherr problem that's been fixed recently - it
> was characterised by compilations in the tcl/unix directory that
> looked like this:
>
> gcc -c -g -O2 -I/star/include -fPIC -I./../generic -I.
> -DPACKAGE_NAME=\"tcl\" -DPACKAGE_TARNAME=\"tcl\"
> -DPACKAGE_VERSION=\"8.2.3\" -DPACKAGE_STRING=\"tcl\ 8.2.3\"
> -DPACKAGE_BUGREPORT=\"[log in to unmask]\" -DPACKAGE=\"tcl\"
> -DVERSION=\"8.2.3\" -DTCL_SHLIB_EXT=\".so\" ./../unix/tclMtherr.c
>
>
> instead of like this:
>
> cc -c -O -D__NO_STRING_INLINES -D__NO_MATH_INLINES -fPIC
> -I./../generic -I. -DHAVE_UNISTD_H=1 -DHAVE_LIMITS_H=1 -DHAVE_GETCWD=1
> -DHAVE_OPENDIR=1 -DHAVE_STRSTR=1 -DHAVE_STRTOL=1 -DHAVE_TMPNAM=1
> -DHAVE_WAITPID=1 -DHAVE_UNISTD_H=1 -DHAVE_SYS_PARAM_H=1 -DUSE_TERMIOS=1
> -DHAVE_SYS_TIME_H=1 -DTIME_WITH_SYS_TIME=1 -DHAVE_TM_ZONE=1
> -DHAVE_TM_GMTOFF=1 -DHAVE_TIMEZONE_VAR=1 -DHAVE_ST_BLKSIZE=1
> -DSTDC_HEADERS=1 -DNEED_MATHERR=1 -DRETSIGTYPE=void -DHAVE_SIGNED_CHAR=1
> -DHAVE_SYS_IOCTL_H=1 -DTCL_SHLIB_EXT=\".so\" ./../unix/tclMtherr.c`
>
> However I get the impression that this is not the one Peter is talking
> about.
That's right. The above was just a fubar by me, the matherr inclusion
problem is a just a repeated symbol that Cygwin and OS X (apparently)
don't like having to deal with. Later versions of Tcl don't have this code
at all (and undefining it in this instance is useless). In general matherr
is deprecated in UNIX and we could actually just remove all the occurances
of it.
Peter.
|