Hi,
In building perl5.001m for SunOS 4.1.3_U1, AIX 3.2.5 and NeXTSTEP 3.2, I've run
across several problems/bugs.
1) After you build perl5 for one architecture, when you run make clean so that
you can build it for a second architecture, the make dies and complains that
the file 'lib/auto/DynaLoader/DynaLoader.a' is not in the right format. It
appears that 'make clean' doesn't get this one. I haven't had a chance to
track it down but it is a problem.
2) There are still some places in perl5 that refer to the 'h2ph' program needing
to be run, however, 'installperl' does not install this. Here is the diff
for how I fixed it:
--------------------------------------------
diff installperl-5.001 installperl
15c15
< @scripts = ('cppstdin', 'c2ph', 'h2xs', 'pstruct', 'x2p/s2p', 'x2p/find2perl',
---
> @scripts = ('cppstdin', 'c2ph', 'h2ph', 'h2xs', 'pstruct', 'x2p/s2p', 'x2p/find2perl',
--------------------------------------------
The alternative would be to update the message(s), wherever they occur.
3) There is a bug in one of the tests for perl. I ran into this when I tried to
verify perl5 December 31, 1994. Since CUT was on January 1, 1995, this bug
appeared. Basically, the file in question is taking the difference between
the CUT date/time stamp and the local date/time stamp. Here is the change I
made to fix the bug:
--------------------------------------------
diff t/op/time.t-5.001 t/op/time.t
44c44
< if (index(" :0:1:-1:365:366:-365:-366:",':' . ($localyday - $yday) . ':') > 0)
---
> if (index(" :0:1:-1:365:364:-365:-364:",':' . ($localyday - $yday) . ':') > 0)
--------------------------------------------
Basically, since what is being compared is a difference in days, the possible
values are 0, +/-1, +/-364 (for a normal year), and +/-365 (for a leap year).
The bug is that the test assumed a normal year of 366 days and a leap year of
367 days.
Please incorporate these changes or better ones if you have them into the next
release of perl5. Thank you.
david@ictv.com
In building perl5.001m for SunOS 4.1.3_U1, AIX 3.2.5 and NeXTSTEP 3.2, I've run
across several problems/bugs.
1) After you build perl5 for one architecture, when you run make clean so that
you can build it for a second architecture, the make dies and complains that
the file 'lib/auto/DynaLoader/DynaLoader.a' is not in the right format. It
appears that 'make clean' doesn't get this one. I haven't had a chance to
track it down but it is a problem.
2) There are still some places in perl5 that refer to the 'h2ph' program needing
to be run, however, 'installperl' does not install this. Here is the diff
for how I fixed it:
--------------------------------------------
diff installperl-5.001 installperl
15c15
< @scripts = ('cppstdin', 'c2ph', 'h2xs', 'pstruct', 'x2p/s2p', 'x2p/find2perl',
---
> @scripts = ('cppstdin', 'c2ph', 'h2ph', 'h2xs', 'pstruct', 'x2p/s2p', 'x2p/find2perl',
--------------------------------------------
The alternative would be to update the message(s), wherever they occur.
3) There is a bug in one of the tests for perl. I ran into this when I tried to
verify perl5 December 31, 1994. Since CUT was on January 1, 1995, this bug
appeared. Basically, the file in question is taking the difference between
the CUT date/time stamp and the local date/time stamp. Here is the change I
made to fix the bug:
--------------------------------------------
diff t/op/time.t-5.001 t/op/time.t
44c44
< if (index(" :0:1:-1:365:366:-365:-366:",':' . ($localyday - $yday) . ':') > 0)
---
> if (index(" :0:1:-1:365:364:-365:-364:",':' . ($localyday - $yday) . ':') > 0)
--------------------------------------------
Basically, since what is being compared is a difference in days, the possible
values are 0, +/-1, +/-364 (for a normal year), and +/-365 (for a leap year).
The bug is that the test assumed a normal year of 366 days and a leap year of
367 days.
Please incorporate these changes or better ones if you have them into the next
release of perl5. Thank you.
david@ictv.com