tom@ICASE.EDU (Tom Crockett) (09/19/90)
Hi,
We're having trouble building Andrew (w/ patches 1-6) under SunOS 4.1 on
a Sun 4/60. We get the following error when doing a make dependInstall
for atk/console/cmd:
doload: Errors while processing
doindex: indexing consolea.do ...doload: Internal botch - should
have resolved in doload_preset
doindex: NULL entrypoint from consolea.do...program terminated!
*** Error code 1
make: Fatal error: Command failed for target `consolea.do'
Current working directory
/usr/local/share/X11R4/src/contrib/toolkits/andrew/atk/console/cmd
*** Error code 1
make: Fatal error: Command failed for target `dependInstall'
Our site.mcr file looks like:
XBASEDIR = /usr/local/X11R4
BASEDIR = /usr/local/X11R4/andrew.006
CDEBUGFLAGS = -pipe
XUTILDIR = $(XBASEDIR)/bin
XINCDIR = $(XBASEDIR)/include/X11
XMKFONTDIR = $(XUTILDIR)/mkfontdir
XLIBDIR = $(XBASEDIR)/lib
XFC = $(XUTILDIR)/bdftosnf
RESOLVLIB = /usr/lib/libresolv.a
#define ConstructMFLAGS
Has anyone else seen this problem? Is there a fix? To get this far, we
also had to patch up something to do with password handling, I forget
the file. Does anyone have a set of patches for SunOS 4.1? The same
configuration builds cleanly under SunOS 4.0.3.
Thanks for any suggestions,
Tom Crockett
ICASE
Institute for Computer Applications in Science and Engineering
M.S. 132C e-mail: tom@icase.edu
NASA Langley Research Center phone: (804) 864-2182
Hampton, VA 23665-5225
tom@ICASE.EDU (Tom Crockett) (09/25/90)
Well, almost a week has gone by and I haven't gotten any responses at all to my message about trouble building Andrew under SunOS 4.1: Excerpts from internet.info-andrew: 19-Sep-90 Trouble Building Andrew Und.. Tom Crockett@icase.edu (1504+0) > We're having trouble building Andrew (w/ patches 1-6) under SunOS 4.1 on > a Sun 4/60. We get the following error when doing a make dependInstall > for atk/console/cmd: > doload: Errors while processing > doindex: indexing consolea.do ...doload: Internal botch - should > have resolved in doload_preset > doindex: NULL entrypoint from consolea.do...program terminated! > *** Error code 1 > make: Fatal error: Command failed for target `consolea.do' > Current working directory > /usr/local/share/X11R4/src/contrib/toolkits/andrew/atk/console/cmd > *** Error code 1 > make: Fatal error: Command failed for target `dependInstall' Should I infer from this that (1) other sites are not seeing this problem, or (2) this is something simple and we just haven't figured it out yet, or (3) nobody is running Andrew under SunOS 4.1? Our sources are current through patchlevel 6, and the same configuration builds and runs cleanly under SunOS 4.0.3, except for minor fixes in overhead/util/lib/getcpw.c and overhead/util/lib/getvpwkey.c. In these two routines, we eliminated references to the pw_quota field of the passwd structure, which, following the System V conventions, is no longer defined in 4.1. We did this by setting STD_DEFINES = -DSUNOS41 in site.mcr and then adding SUNOS41 to the #if tests for systems which don't support pw_quota. Our assumption is that this symbol is not currently being used elsewhere in the sources, and so should be side-effect free (and even if it were defined elsewhere, it's what we would want to do anyway). Is this a poor assumption? Tom Crockett ICASE Institute for Computer Applications in Science and Engineering M.S. 132C e-mail: tom@icase.edu NASA Langley Research Center phone: (804) 864-2182 Hampton, VA 23665-5225
guy@auspex.auspex.com (Guy Harris) (09/27/90)
>In these two routines, we eliminated references to the pw_quota field of the >passwd structure, which, following the System V conventions, is no >longer defined in 4.1. The reason it was done is that password aging (S5-style) is supported by SunOS 4.1, which means that in both environments (S5 and BSD) you want to provide a "pw_age" field in the "struct passwd" structure. The BSD "getpw*" routines, and the SunOS 4.0[.x] BSD-environment routines, *always* set the "pw_quota" field to zero, so... >We did this by setting STD_DEFINES = -DSUNOS41 in site.mcr and then >adding SUNOS41 to the #if tests for systems which don't support pw_quota. >Our assumption is that this symbol is not currently being used elsewhere >in the sources, ...if somebody *is* using it, I'm not sure why. The two routines you list (BTW, it's "gtvpwkey.c", not "getvpwkey.c", at least in the source we have here) set it to zero as well.