Not OK: perl 5.00562 on os390 05.00 (UNINSTALLED) #803
Comments
From pvhp@forte.comHere is a summary of the failures seen during `make test` op/pack.............FAILED at test 152 op/regexp...........CEE5213S The signal SIGPIPE was received. pragma/locale.......CEE3703I In HPCB Control Block, the Eye Catcher is damaged. pragma/utf8.........Malformed UTF-8 character at pragma/utf8.t line 22. lib/bigfloat........CEE5213S The signal SIGPIPE was received. lib/charnames.......CEE5213S The signal SIGPIPE was received. lib/dumper..........CEE5213S The signal SIGPIPE was received. lib/io_unix.........Can't call method "getline" on an undefined value at Failed 12 test scripts out of 217, 94.47% okay. Perl Info |
From @jhi
The "Eye Catcher is damaged"? Who said mainframe error messages are boring? -- |
From [Unknown Contact. See original ticket]Jarkko Hietaniemi wrote:
Indeed. BTW I note that ord("!") == 90 here so I can get around the one failure Inline Patch--- lib/unicode/Name.pl.orig Tue Nov 2 16:36:49 1999
+++ lib/unicode/Name.pl Tue Nov 2 16:37:36 1999
@@ -1,7 +1,7 @@
return <<'END';
0000 001f <control>
0020 SPACE
-0021 EXCLAMATION MARK
+005a EXCLAMATION MARK
0022 QUOTATION MARK
0023 NUMBER SIGN
0024 DOLLAR SIGN
but obviously I cannot impose such changes on the ascii world so Peter Prymmer |
From @jhiPeter Prymmer writes:
You cannot impose such changes on the *Unicode* world, HTH.
There is a paper at the Unicode web page about EBCDIC. Somebody who -- |
From [Unknown Contact. See original ticket]The enclosed patch ought to take care of the following failures op/pack.............FAILED at test 152 Most of these fixes are simple encoded character set issues. For The failures in t/pragma/locale.t clearly indicate two broken locales % ls -1 /usr/lib/nls/locale | wc -l locales, including 44 unique encodings such as "IBM-1149@euro", but I This patch does not address the following failures that were pragma/utf8.........FAILED at test 0 I suspect there are some lingering asciiisms in toke.c perhaps sv.c lib/bigfloat........FAILED at test 38 These two look nasty since they look like perl is having trouble lib/charnames.......FAILED at test 1 There are at least two ways to fix this one on ebcdic platforms. lib/io_unix.........FAILED at test 3 This one appears to result from C<$sock->accept();> returning an Alright here is the unified diff (which must be applied with a -u begin 644 62.patch.gz Peter Prymmer |
From @jhiPeter Prymmer writes:
Excellent. Big Iron just keeps going...
I must ask: how clearly? What goes wrong? Which tests fail and how?
Very prudent.
That would be me. I'll try to grab a tuit.
Yes, please.
-- |
From [Unknown Contact. See original ticket]Jarkko Hietaniemi wrote in response to me:
I am glad that you asked since in going back over things ... it seems to Specifically the first call to: setlocale(LC_ALL, "th_th.ISO8859-11") In clear text my "correction" to locale.t was to merely delete Svenska --- t/pragma/locale.t.orig Wed Nov 10 10:21:26 1999 +if ($^O eq 'os390') { And I just noted that I might be able to get away with leaving the Here is the way I uncovered the problem: if I revert to the original sub trylocale { Then when I run the test like so: ./perl -T t/pragma/locale.t.diag So apparently C<setlocale(LC_ALL, "th_th.ISO8859-11")> should not I have further tried to narrow things down. If I shorten the list of my $locales = <<EOF; And I redo the diagnostics around setlocale() like so: sub trylocale { then I can see the following: ./perl -T t/pragma/locale.t.new I have also been able to tickle Perl's C<Out of Memory!> error during I did carried out all the tests described above on an OE R 2.5 system.
Will do in a separate message. Peter Prymmer |
From [Unknown Contact. See original ticket]Jarkko Hietaniemi wrote Wed, 3 Nov 1999:
OK, If the intent of charnames is to do _only_ a unicode-ish encoding Inline Patch--- t/lib/charnames.t.orig Tue Nov 2 16:08:56 1999
+++ t/lib/charnames.t Mon Nov 8 12:06:05 1999
@@ -12,7 +12,7 @@
use charnames ':full';
-print "not " unless "Here\N{EXCLAMATION MARK}?" eq 'Here!?';
+print "not " unless "Here\N{EXCLAMATION MARK}?" eq "Here\041?";
print "ok 1\n";
print "# \$res=$res \$\@='$@'\nnot "
However, if the intent was to have a script like: use charnames ':full'; print out 'yes!' then the following patch would be necessary, in fact I Here is the one I prefer. Unfortunately, when I run the unmodified % ./perl t/lib/charnames.t.orig where the "Attempt..." message is generated right after the execution of Inline Patch--- lib/unicode/Name.pl.orig Fri Nov 5 14:58:03 1999
+++ lib/unicode/Name.pl Mon Nov 8 12:08:56 1999
@@ -1,4 +1,5 @@
-return <<'END';
+package charnames;
+my $table = <<'END';
0000 001f <control>
0020 SPACE
0021 EXCLAMATION MARK
@@ -10547,3 +10548,247 @@
fffc OBJECT REPLACEMENT CHARACTER
fffd REPLACEMENT CHARACTER
END
+
+# are we on an EBCDIC platform?
+
+if ($^O eq 'os390' || $^O eq 'vmesa' || $^O eq 'os400' || $^O eq 'posix-bc') {
+
+ # remove iso-latin chrs 0000..00ff
+
+ $table =~ s/.+(0100\t)/$1/s;
+
+ # prepend ebcdic chrs from codepage 1047
+
+ $table = <<"END1047";
+0000 003f <control>
+0040 SPACE
+005a EXCLAMATION MARK
+007f QUOTATION MARK
+007b NUMBER SIGN
+005b DOLLAR SIGN
+006c PERCENT SIGN
+0050 AMPERSAND
+007d APOSTROPHE
+004d LEFT PARENTHESIS
+005d RIGHT PARENTHESIS
+005c ASTERISK
+004e PLUS SIGN
+006b COMMA
+0060 HYPHEN-MINUS
+004b FULL STOP
+0061 SOLIDUS
+00f0 DIGIT ZERO
+00f1 DIGIT ONE
+00f2 DIGIT TWO
+00f3 DIGIT THREE
+00f4 DIGIT FOUR
+00f5 DIGIT FIVE
+00f6 DIGIT SIX
+00f7 DIGIT SEVEN
+00f8 DIGIT EIGHT
+00f9 DIGIT NINE
+007a COLON
+005e SEMICOLON
+004c LESS-THAN SIGN
+007e EQUALS SIGN
+006e GREATER-THAN SIGN
+006f QUESTION MARK
+007c COMMERCIAL AT
+00c1 LATIN CAPITAL LETTER A
+00c2 LATIN CAPITAL LETTER B
+00c3 LATIN CAPITAL LETTER C
+00c4 LATIN CAPITAL LETTER D
+00c5 LATIN CAPITAL LETTER E
+00c6 LATIN CAPITAL LETTER F
+00c7 LATIN CAPITAL LETTER G
+00c8 LATIN CAPITAL LETTER H
+00c9 LATIN CAPITAL LETTER I
+00d1 LATIN CAPITAL LETTER J
+00d2 LATIN CAPITAL LETTER K
+00d3 LATIN CAPITAL LETTER L
+00d4 LATIN CAPITAL LETTER M
+00d5 LATIN CAPITAL LETTER N
+00d6 LATIN CAPITAL LETTER O
+00d7 LATIN CAPITAL LETTER P
+00d8 LATIN CAPITAL LETTER Q
+00d9 LATIN CAPITAL LETTER R
+00e2 LATIN CAPITAL LETTER S
+00e3 LATIN CAPITAL LETTER T
+00e4 LATIN CAPITAL LETTER U
+00e5 LATIN CAPITAL LETTER V
+00e6 LATIN CAPITAL LETTER W
+00e7 LATIN CAPITAL LETTER X
+00e8 LATIN CAPITAL LETTER Y
+00e9 LATIN CAPITAL LETTER Z
+00ad LEFT SQUARE BRACKET
+00e0 REVERSE SOLIDUS
+00bd RIGHT SQUARE BRACKET
+005f CIRCUMFLEX ACCENT
+006d LOW LINE
+0079 GRAVE ACCENT
+0081 LATIN SMALL LETTER A
+0082 LATIN SMALL LETTER B
+0083 LATIN SMALL LETTER C
+0084 LATIN SMALL LETTER D
+0085 LATIN SMALL LETTER E
+0086 LATIN SMALL LETTER F
+0087 LATIN SMALL LETTER G
+0088 LATIN SMALL LETTER H
+0089 LATIN SMALL LETTER I
+0091 LATIN SMALL LETTER J
+0092 LATIN SMALL LETTER K
+0093 LATIN SMALL LETTER L
+0094 LATIN SMALL LETTER M
+0095 LATIN SMALL LETTER N
+0096 LATIN SMALL LETTER O
+0097 LATIN SMALL LETTER P
+0098 LATIN SMALL LETTER Q
+0099 LATIN SMALL LETTER R
+00a2 LATIN SMALL LETTER S
+00a3 LATIN SMALL LETTER T
+00a4 LATIN SMALL LETTER U
+00a5 LATIN SMALL LETTER V
+00a6 LATIN SMALL LETTER W
+00a7 LATIN SMALL LETTER X
+00a8 LATIN SMALL LETTER Y
+00a9 LATIN SMALL LETTER Z
+00c0 LEFT CURLY BRACKET
+004f VERTICAL LINE
+00d0 RIGHT CURLY BRACKET
+00a1 TILDE
+00ff 00ff <control>
+0041 NO-BREAK SPACE
+00aa INVERTED EXCLAMATION MARK
+004a CENT SIGN
+00b1 POUND SIGN
+009f CURRENCY SIGN
+00b2 YEN SIGN
+006a BROKEN BAR
+00b5 SECTION SIGN
+00bb DIAERESIS
+00b4 COPYRIGHT SIGN
+009a FEMININE ORDINAL INDICATOR
+008a LEFT-POINTING DOUBLE ANGLE QUOTATION MARK
+00b0 NOT SIGN
+00ca SOFT HYPHEN
+00af REGISTERED SIGN
+00bc MACRON
+0090 DEGREE SIGN
+008f PLUS-MINUS SIGN
+00ea SUPERSCRIPT TWO
+00fa SUPERSCRIPT THREE
+00be ACUTE ACCENT
+00a0 MICRO SIGN
+00b6 PILCROW SIGN
+00b3 MIDDLE DOT
+009d CEDILLA
+00da SUPERSCRIPT ONE
+009b MASCULINE ORDINAL INDICATOR
+008b RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK
+00b7 VULGAR FRACTION ONE QUARTER
+00b8 VULGAR FRACTION ONE HALF
+00b9 VULGAR FRACTION THREE QUARTERS
+00ab INVERTED QUESTION MARK
+0064 LATIN CAPITAL LETTER A WITH GRAVE
+0065 LATIN CAPITAL LETTER A WITH ACUTE
+0062 LATIN CAPITAL LETTER A WITH CIRCUMFLEX
+0066 LATIN CAPITAL LETTER A WITH TILDE
+0063 LATIN CAPITAL LETTER A WITH DIAERESIS
+0067 LATIN CAPITAL LETTER A WITH RING ABOVE
+009e LATIN CAPITAL LETTER AE
+0068 LATIN CAPITAL LETTER C WITH CEDILLA
+0074 LATIN CAPITAL LETTER E WITH GRAVE
+0071 LATIN CAPITAL LETTER E WITH ACUTE
+0072 LATIN CAPITAL LETTER E WITH CIRCUMFLEX
+0073 LATIN CAPITAL LETTER E WITH DIAERESIS
+0078 LATIN CAPITAL LETTER I WITH GRAVE
+0075 LATIN CAPITAL LETTER I WITH ACUTE
+0076 LATIN CAPITAL LETTER I WITH CIRCUMFLEX
+0077 LATIN CAPITAL LETTER I WITH DIAERESIS
+00ac LATIN CAPITAL LETTER ETH
+0069 LATIN CAPITAL LETTER N WITH TILDE
+00ed LATIN CAPITAL LETTER O WITH GRAVE
+00ee LATIN CAPITAL LETTER O WITH ACUTE
+00eb LATIN CAPITAL LETTER O WITH CIRCUMFLEX
+00ef LATIN CAPITAL LETTER O WITH TILDE
+00ec LATIN CAPITAL LETTER O WITH DIAERESIS
+00bf MULTIPLICATION SIGN
+0080 LATIN CAPITAL LETTER O WITH STROKE
+00fd LATIN CAPITAL LETTER U WITH GRAVE
+00fe LATIN CAPITAL LETTER U WITH ACUTE
+00fb LATIN CAPITAL LETTER U WITH CIRCUMFLEX
+00fc LATIN CAPITAL LETTER U WITH DIAERESIS
+00ba LATIN CAPITAL LETTER Y WITH ACUTE
+00ae LATIN CAPITAL LETTER THORN
+0059 LATIN SMALL LETTER SHARP S
+0044 LATIN SMALL LETTER A WITH GRAVE
+0045 LATIN SMALL LETTER A WITH ACUTE
+0042 LATIN SMALL LETTER A WITH CIRCUMFLEX
+0046 LATIN SMALL LETTER A WITH TILDE
+0043 LATIN SMALL LETTER A WITH DIAERESIS
+0047 LATIN SMALL LETTER A WITH RING ABOVE
+009c LATIN SMALL LETTER AE
+0048 LATIN SMALL LETTER C WITH CEDILLA
+0054 LATIN SMALL LETTER E WITH GRAVE
+0051 LATIN SMALL LETTER E WITH ACUTE
+0052 LATIN SMALL LETTER E WITH CIRCUMFLEX
+0053 LATIN SMALL LETTER E WITH DIAERESIS
+0058 LATIN SMALL LETTER I WITH GRAVE
+0055 LATIN SMALL LETTER I WITH ACUTE
+0056 LATIN SMALL LETTER I WITH CIRCUMFLEX
+0057 LATIN SMALL LETTER I WITH DIAERESIS
+008c LATIN SMALL LETTER ETH
+0049 LATIN SMALL LETTER N WITH TILDE
+00cd LATIN SMALL LETTER O WITH GRAVE
+00ce LATIN SMALL LETTER O WITH ACUTE
+00cb LATIN SMALL LETTER O WITH CIRCUMFLEX
+00cf LATIN SMALL LETTER O WITH TILDE
+00cc LATIN SMALL LETTER O WITH DIAERESIS
+00e1 DIVISION SIGN
+0070 LATIN SMALL LETTER O WITH STROKE
+00dd LATIN SMALL LETTER U WITH GRAVE
+00de LATIN SMALL LETTER U WITH ACUTE
+00db LATIN SMALL LETTER U WITH CIRCUMFLEX
+00dc LATIN SMALL LETTER U WITH DIAERESIS
+008d LATIN SMALL LETTER Y WITH ACUTE
+008e LATIN SMALL LETTER THORN
+00df LATIN SMALL LETTER Y WITH DIAERESIS
+$table
+END1047
+
+ chomp($table);
+
+ if ($^O eq 'os400') {
+ # codepage 1047 -> 819
+ $table =~ s/005f( CIRCUMFLEX ACCENT)/00b0$1/;
+ $table =~ s/00ad( LEFT SQUARE BRACKET)/00ba$1/;
+ $table =~ s/00b0( NOT SIGN)/005f$1/;
+ $table =~ s/00ba( LATIN CAPITAL LETTER Y WITH ACUTE)/00ad$1/;
+ $table =~ s/00bb( DIAERESIS)/00bd$1/;
+ $table =~ s/00bd( RIGHT SQUARE BRACKET)/00bb$1/;
+ }
+
+ if ($^O eq 'posix-bc') {
+ # codepage 1047 -> posix-bc
+ $table =~ s/00ad( LEFT SQUARE BRACKET)/00bb$1/;
+ $table =~ s/00e0( REVERSE SOLIDUS)/00bc$1/;
+ $table =~ s/005f( CIRCUMFLEX ACCENT)/006a$1/;
+ $table =~ s/0079( GRAVE ACCENT)/004a$1/;
+ $table =~ s/00c0( LEFT CURLY BRACKET)/00fb$1/;
+ $table =~ s/00d0( RIGHT CURLY BRACKET)/00fd$1/;
+ $table =~ s/00a1( TILDE)/00ff$1/;
+ $table =~ s/00ff 00ff( <control>)/005f 005f$1/;
+ $table =~ s/004a( CENT SIGN)/00b0$1/;
+ $table =~ s/006a( BROKEN BAR)/00d0$1/;
+ $table =~ s/00bb( DIAERESIS)/0079$1/;
+ $table =~ s/00b0( NOT SIGN)/00ba$1/;
+ $table =~ s/00bc( MACRON)/00a1$1/;
+ $table =~ s/00fd( LATIN CAPITAL LETTER U WITH GRAVE)/00e0$1/;
+ $table =~ s/00fb( LATIN CAPITAL LETTER U WITH CIRCUMFLEX)/00dd$1/;
+ $table =~ s/00ba( LATIN CAPITAL LETTER Y WITH ACUTE)/00ad$1/;
+ $table =~ s/00dd( LATIN SMALL LETTER U WITH GRAVE)/00c0$1/;
+ }
+}
+
+return $table;
+
Thanks. I've looked through the unicode.org paper on a proposal for I have yet to study perl's C<use utf8;> pragma much though. Even if Peter Prymmer |
From [Unknown Contact. See original ticket]Peter Prymmer <pvhp@forte.com> wrote
Since that message is a "should not happen", surely the thing is to track Mike Guy |
From @jhi: --- lib/unicode/Name.pl.orig Fri Nov 5 14:58:03 1999 After some thought: I'd like this patch go to somewhere else than -- |
From [Unknown Contact. See original ticket]Jarkko Hietaniemi wrote:
[snip]
OK. I suspected that the first reason you list would have had Is it not the case that perl will be trying to allow switching print "hello world\n"; to switch to printing out the unicode (not utf8) version of Peter Prymmer |
From @jhiPeter Prymmer writes:
It's not pretty, yes, and a lot of work, but I do not think littering -- |
From [Unknown Contact. See original ticket]Peter Prymmer writes:
AFAICS, your question is about communication of Perl with the external Ilya |
From [Unknown Contact. See original ticket]Jarkko Hietaniemi wrote:
Yes I agree with you (and am not trying to argue for the patch). Peter Prymmer |
Migrated from rt.perl.org#1730 (status was 'resolved')
Searchable as RT1730$
The text was updated successfully, but these errors were encountered: