develooper Front page | perl.perl5.porters | Postings from August 2000

[PATCH: 6948] add SCNfldbl to configure.com

Thread Next
From:
Peter Prymmer
Date:
August 31, 2000 18:45
Subject:
[PATCH: 6948] add SCNfldbl to configure.com
Message ID:
Pine.OSF.4.10.10008311811060.91659-100000@aspara.forte.com

With the following patch to the 6948 kit I can configure and build perl.  
The 3 test failures I now see with DEC C 6.0 are:

[.lib]ftmp-mktemp.......ok
Error printing to tempfile

[.lib]ftmp-posix........FAILED on test 6

<snip>

[.lib]st-recurse........ok
Corrupted storable file (binary v2.1) at ../../lib/storable.pm (autosplit
into [
--.lib.auto.Storable]retrieve.al) line 198, at [.lib]st-retrieve.t line 51

[.lib]st-retrieve.......FAILED on test 6
Corrupted storable file (binary v2.1) at ../../lib/storable.pm (autosplit
into [
--.lib.auto.Storable]retrieve.al) line 198, at [.lib]st-store.t line 46

[.lib]st-store..........FAILED on test 3
[.lib]st-tied...........ok

<snip>

Failed 3/249 tests, 94.78% okay.
u=23.01  s=0  cu=0  cs=0  files=239  tests=10259

in particular lib/trig.t is now passing ok (much to my happy surprise).

I have tried commenting out the unlink()s from tests like
lib/st-retrieve.t and it appears at first glance that the fix to
vms.c:my_fwrite() that Craig supplied is OK in that the [.t]store. and
[.t]nstore. files left behind have more that just the magic pst0 string in
them now, although I have yet to run DUMP or od on them yet.  Perhaps
the problem there is merely reading from the RMS file(?).

I have checked to see what could be causing the extraneous newlines
in the output of test.com such as shown above, or e.g.:

---------------------------------
[.op]subst_wamp.........ok
[.op]sysio..............ok
# Note: logical name 'PATH' may have been deleted

[.op]taint..............ok
---------------------------------

At one point I reverted my_fwrite() to return 1; to see if that had an
effect.  It did not hence I suspect that Charles Lane's pipe patch is the
culprit behind the messy output of test.com nowadays.

In anticipation of *SCNgldbl,*SCNeldbl variables et alia I do have a
request of our glorious Configure pumpking: Could we please refrain from
things like d_PRIEldbl vs. d_PRIeldbl?  DCL cannot distinguish the case of
shell variables used.  Although munchconfig can handle the difference, the
shell on VMS cannot and I'd also like to mention that neither can cmd.exe
on NT nor MPW on Mac OS (although admittedly those platforms do not
currently pull their config.h files out of config_h.SH).  If you try:

  c:\> echo %COMSPEC% %ComSpec%

or:

  MPW> echo "{HOME} {hOMe}"

you'll be seeing double.  Thank you very much for considering it.

Here's the patch:

--- configure.com.orig	Wed Aug 30 12:34:17 2000
+++ configure.com	Thu Aug 31 18:07:13 2000
@@ -2764,6 +2764,8 @@
 $   d_isnanl = "undef"
 $   d_modfl = "undef"
 $ ENDIF
+$ d_SCNfldbl = d_PRIfldbl
+$ sSCNfldbl = sPRIfldbl ! expect consistency
 $!
 $! Now some that we build up
 $!
@@ -4641,6 +4643,7 @@
 $ WC "d_PRIo64='" + d_PRIo64 + "'"
 $ WC "d_PRIu64='" + d_PRIu64 + "'"
 $ WC "d_PRIx64='" + d_PRIx64 + "'"
+$ WC "d_SCNfldbl='" + d_SCNfldbl + "'"
 $ WC "d_access='" + d_access + "'"
 $ WC "d_accessx='undef'"
 $ WC "d_alarm='define'"
@@ -5120,6 +5123,7 @@
 $ WC "sPRIo64='" + sPRIo64 + "'"
 $ WC "sPRIu64='" + sPRIu64 + "'"
 $ WC "sPRIx64='" + sPRIx64 + "'"
+$ WC "sSCNfldbl='" + sSCNfldbl + "'"
 $ WC "sched_yield='" + sched_yield + "'"
 $ WC "seedfunc='" + seedfunc + "'"
 $ WC "selectminbits='32'"
End of Patch.

Peter Prymmer




Thread Next


nntp.perl.org: Perl Programming lists via nntp and http.
Comments to Ask Bjørn Hansen at ask@perl.org | Group listing | About