Glenn Fowler wrote:
>
> On Wed, 28 Jun 2006 13:19:37 +0200 Roland Mainz wrote:
> > After another hour of digging I realised that I have the new "map.c" in
> > my tree. Removing the patch seems to trigger the problem in the original
> > report - and re-appying the "map.c" change fixes it (except
> > "./src/cmd/ksh93/tests/options.sh[118]: ./-ksh ignores .profile", e.g
> > only "./src/cmd/ksh93/tests/options.sh[99]" and
> > "./src/cmd/ksh93/tests/options.sh[101]" disappear).
>
> I fear you may be testing a frankenstein in some cases
>
> the features/map.c change must be propagated to all binary components
> under test { ksh libshell libast }
> also make sure that LD_LIBRARY_PATH_FLAVOR_OF_THE_DAY is set correctly
> to access the appropriate { libshell libast } .so's
I usually use "rm -Rf built/ ; gunzip source ; patch source ; built it"
for such tests so I hope this avoids the "frankenstein" syndrome... and
for the Solaris OS/Net tree I used
-- snip --
(cd libast ; make clean ; make install_h ; make install)
(cd libdll ; make clean ; make install_h ; make install)
(cd libcmd ; make clean ; make install_h ; make install)
(cd libshell ; make clean ; make install_h ; make install)
(cd ../cmd/ksh ; make clean ; make install)
-- snip --
I think that is sufficient to make sure that all object files and
includes have been updated properly.
> the test does have a conflict with FIGNORE in the environment
> that could change the -G expected results
> this will be fixed
Ok... thanks! :-)
----
Bye,
Roland
--
__ . . __
(o.\ \/ /.o) roland.mainz at nrubsig.org
\__\/\/__/ MPEG specialist, C&&JAVA&&Sun&&Unix programmer
/O /==\ O\ TEL +49 641 7950090
(;O/ \/ \O;)
