PW32: Regression testing results for ported software

Common fail reasons

EXT Dependence on external non-PW32 application, which doesn't function as Unix one should. Specific errors:
EXT_CRLF
External programs outputs \r\n-lineended files
PERM Full permission mapping not implemented
NDET Nondeterministic fail/pass result
SUPPOSED Tests which explicitly stated as allowed to fail under some circumstances

Current status

Package
Test Description
autoconf 2.14a CVS snapshot 20000528 (97/98)
#65 (macros.m4:50)Fails due to lack of X headers/libraries
automake 1.4a-dep CVS snapshot 20000528 (249/249)
*two tests required non-test-functionality related patches:
ansi3.testneeded copy files instead of symlink, because gcc doesn't understand them
pr19.test needed to disable automatical dependency tracking, since gcc output filenames in DOS format, on which make chokes.
bash 2.03
run-execscript Gives errors related to execing / , let it wait in peace
run-jobs bash 2.03 has bugs which make it fail to build if job control is not enabled. Of course pw32 lacks features for it to be functional.
run-new-expSUPPOSED
as test itself warns, it's supposed to fail with pw32. It however just hang in cat.
run-shopt barfs due to lack of readline
run-testPERM
fails due to lack of full special devices support (/dev/tty should be character), permission mapping (tests setuid/setgid), /dev/fd filesystem
run-trapSUPPOSED
different signal mapping
diffutils 2.7.2
allPASSED
gawk 3.04 (87/88)
pidFails due to extra proxy parent for each exec'ed child
grep 2.0
allPASSED
grep 2.4.2 (8/8)
allPASSED
gzip 1.2.4
allPASSED
flex 2.5.4a
allPASSED
m4 1.4 (57/57)
allPASSED
sed 3.02
allPASSED
sh-utils 1.16
niceDJGPP nice seems to be implemented (well, faked), the other way.
tar 1.12
incremen.shNDET
AFAIU reason for nondeterministic failing is 2sec-resolution of times on FAT filesystem. Putting sleep in the appropriate place makes it pass robustly.
textutils 1.22f
allPASSED

Paul Sokolovsky | use this form to submit bug