FileCache only works in "main" package #7693
Comments
From @kaibolayThis is a bug report for perl from kai@bolay.de, ----------------------------------------------------------------- I As a workaround I temporarily switch to the "main" package when using $ cat FileCacheBug.pl use strict; eval { eval { no strict "refs"; if (!$@) { eval { no strict "refs"; if (!$@) { $ ./FileCacheBug.pl Flags: This perlbug was built using Perl v5.8.5 in the Red Hat build system. Site configuration information for perl v5.8.5: Configured by Red Hat, Inc. at Tue Oct 12 12:44:18 EDT 2004. Summary of my perl5 (revision 5 version 8 subversion 5) configuration: Locally applied patches: @INC for perl v5.8.5: Environment for perl v5.8.5: |
From @schwernOn Tue, Dec 07, 2004 at 10:45:16PM -0000, kai @ bolay. de wrote:
Unless I'm misunderstanding, this is not a bug in FileCache. This is how use Text::Soundex; package Foo; F000 Module will typically only export their functions into the current package. If you want to use FileCache or any module's functions in a package you
Add a "use File::Cache;" here and you're good.
-- |
|
The RT System itself - Status changed from 'new' to 'open' |
From @tamiasOn Wed, Dec 08, 2004 at 12:06:29AM -0500, Michael G Schwern wrote:
I was going to make the same suggestion, but I tried it, and it didn't Ronald |
From @schwernOn Wed, Dec 08, 2004 at 12:38:33AM -0500, Ronald J Kimball wrote:
Its line 1. The problem appears to be that its exporting to caller(1) not caller(0). -- |
From kane@xs4all.netOn Dec 8, 2004, at 7:09 AM, Michael G Schwern wrote:
I'll send a patch for use of Exporter + tests along this afternoon -- 'Real programmers use "cat > a.out"' CPANPLUS http://cpanplus.sf.net |
From kane@xs4all.netOn Dec 8, 2004, at 9:57 AM, Jos I. Boumans wrote:
Patch attached to make FileCache use Exporter, plus a new test file lib/FileCache/t/01open....................ok -- From kid's Superman costume for Halloween (stitched into the cape's CPANPLUS http://cpanplus.sf.net |
From kane@xs4all.netfilecache-exporter-[#32949].patch--- MANIFEST.org Wed Dec 8 11:02:20 2004
+++ MANIFEST Wed Dec 8 10:37:27 2004
@@ -1345,6 +1345,7 @@
lib/FileCache/t/03append.t See if FileCache works
lib/FileCache/t/04twoarg.t See if FileCache works
lib/FileCache/t/05override.t See if FileCache works
+lib/FileCache/t/06export.t See if FileCache exporting works
lib/File/CheckTree.pm Perl module supporting wholesale file mode validation
lib/File/CheckTree.t See if File::CheckTree works
lib/File/Compare.pm Emulation of cmp command
--- lib/FileCache.pm.org Wed Dec 8 10:10:12 2004
+++ lib/FileCache.pm Wed Dec 8 14:20:33 2004
@@ -84,15 +84,27 @@
# These are not C<my> for legacy reasons.
# Previous versions requested the user set $cacheout_maxopen by hand.
# Some authors fiddled with %saw to overcome the clobber on initial open.
-use vars qw(%saw $cacheout_maxopen);
+use vars qw(%saw $cacheout_maxopen @EXPORT);
my %isopen;
my $cacheout_seq = 0;
sub import {
my ($pkg,%args) = @_;
- $pkg = caller(1);
- *{$pkg.'::cacheout'} = \&cacheout;
- *{$pkg.'::close'} = \&cacheout_close;
+
+ # Not using Exporter is naughty.
+ # Also, using caller(1) is just wrong.
+ #$pkg = caller(1);
+ #*{$pkg.'::cacheout'} = \&cacheout;
+ #*{$pkg.'::close'} = \&cacheout_close;
+
+ # Use Exporter. %args are for us, not Exporter.
+ # Make sure to up export_to_level, or we will import into ourselves,
+ # rather than our calling package;
+ use base 'Exporter';
+ @EXPORT = qw[cacheout cacheout_close];
+
+ __PACKAGE__->export_to_level(1);
+ Exporter::import( $pkg );
# Truth is okay here because setting maxopen to 0 would be bad
return $cacheout_maxopen = $args{maxopen} if $args{maxopen};
diff -u /dev/null lib/FileCache/t/06export.t
--- /dev/null Wed Dec 8 11:05:55 2004
+++ lib/FileCache/t/06export.t Wed Dec 8 11:05:39 2004
@@ -0,0 +1,62 @@
+#!./perl
+BEGIN {
+ chdir 't' if -d 't';
+
+ #For tests within the perl distribution
+ @INC = '../lib' if -d '../lib';
+ END;
+
+ # Functions exported by FileCache;
+ @funcs = qw[cacheout cacheout_close];
+ $i = 0;
+
+ # number of tests
+ print "1..8\n";
+}
+
+# Test 6: Test that exporting both works to package main and
+# other packages. Now using Exporter.
+
+# First, we shouldn't be able to have these in our namespace
+# Add them to BEGIN so the later 'use' doesn't influence this
+# test
+BEGIN {
+ for my $f (@funcs) {
+ ++$i;
+ print 'not ' if __PACKAGE__->can($f);
+ print "ok $i\n";
+ }
+}
+
+# With an empty import list, we also shouldn't have them in
+# our namespace.
+# Add them to BEGIN so the later 'use' doesn't influence this
+# test
+BEGIN {
+ use FileCache ();
+ for my $f (@funcs) {
+ ++$i;
+ print 'not ' if __PACKAGE__->can($f);
+ print "ok $i\n";
+ }
+}
+
+
+# Now, we use FileCache in 'main'
+{ use FileCache;
+ for my $f (@funcs) {
+ ++$i;
+ print 'not ' if !__PACKAGE__->can($f);
+ print "ok $i\n";
+ }
+}
+
+# Now we use them in another package
+{ package X;
+ use FileCache;
+ for my $f (@main::funcs) {
+ ++$main::i;
+ print 'not ' if !__PACKAGE__->can($f);
+ print "ok $main::i\n";
+ }
+}
|
From @rgsJos I. Boumans wrote:
Thanks, applied as #23627. |
|
@rgs - Status changed from 'open' to 'resolved' |
|
@kaibolay - Status changed from 'resolved' to 'open' |
From @kaibolayThanks for the quick response(s)! Unfortunately the socond aspect of the bug is still there. I think it is # Open in their package. I attached a patch to your 06export.t which tries to show the problem. Here's the output when I run the patched test script with your patched 1..11 I guess the problems on line 162 and 165 in FileCache.pm should also be Thanks so much! |
From @kaibolaybug.patch*** 06export.t.org 2004-12-08 15:59:10.005377914 -0500
--- 06export.t 2004-12-08 15:56:34.527858529 -0500
***************
*** 1,4 ****
! #!./perl
BEGIN {
chdir 't' if -d 't';
--- 1,4 ----
! #!./perl -w
BEGIN {
chdir 't' if -d 't';
*************** BEGIN {
*** 8,17 ****
# Functions exported by FileCache;
@funcs = qw[cacheout cacheout_close];
$i = 0;
# number of tests
! print "1..8\n";
}
# Test 6: Test that exporting both works to package main and
--- 8,18 ----
# Functions exported by FileCache;
@funcs = qw[cacheout cacheout_close];
+
$i = 0;
# number of tests
! print "1..11\n";
}
# Test 6: Test that exporting both works to package main and
*************** BEGIN {
*** 39,44 ****
--- 40,54 ----
print 'not ' if __PACKAGE__->can($f);
print "ok $i\n";
}
+
+ ++$i;
+ my $file = "foo.$$.".time();
+ FileCache::cacheout $file; print $file "content"; close($file);
+ my $content;
+ if (open($file, "<", $file)) {local $/; $content=<$file>; close ($file);}
+ unlink($file);
+ print "not " if ($content ne "content");
+ print "ok $i\n";
}
*************** BEGIN {
*** 49,54 ****
--- 59,73 ----
print 'not ' if !__PACKAGE__->can($f);
print "ok $i\n";
}
+
+ ++$i;
+ my $file = "foo.$$.".time();
+ cacheout $file; print $file "content"; close($file);
+ my $content;
+ if (open($file, "<", $file)) {local $/; $content=<$file>; close ($file);}
+ unlink($file);
+ print "not " if ($content ne "content");
+ print "ok $i\n";
}
# Now we use them in another package
*************** BEGIN {
*** 59,62 ****
--- 78,90 ----
print 'not ' if !__PACKAGE__->can($f);
print "ok $main::i\n";
}
+
+ ++$main::i;
+ my $file = "foo.$$.".time();
+ cacheout $file; print $file "content"; close($file);
+ my $content;
+ if (open($file, "<", $file)) {local $/; $content=<$file>; close ($file);}
+ unlink($file);
+ print "not " if ($content ne "content");
+ print "ok $main::i\n";
}
|
From @schwern
This much at least is because FileCache::import has not yet been called When you do "use FileCache ()" import is not called so $cache_maxopen is |
From @schwern
The rest of this is because, I think, FileCache does not check to see if In your test you do this. cacheout $file; FileCache does not check to see if the file has gone away and try to This is separate from the exporting problem. |
From @kaibolayMichael G Schwern via RT wrote:
I think you're misreading my test script. The second (and third) What I do is: { my $file = "filename"; { my $file = "filename"; [...]
I don't think so. I think the "print() on unopened filehandle" is caused sub cacheout_open { Jos I. Boumans <kane@xs4all.net> replaced similar uses of caller(1) in I've updated my test script to use $file1, $file2, and $file3 as [kbolay@dose FileCacheBug]$ cat ./06export.t #For tests within the perl distribution # Functions exported by FileCache; $i = 0; # number of tests # Test 6: Test that exporting both works to package main and # First, we shouldn't be able to have these in our namespace # With an empty import list, we also shouldn't have them in ++$i; # Now, we use FileCache in 'main' ++$i; # Now we use them in another package ++$main::i; [kbolay@dose FileCacheBug]$ perl ./06export.t |
From @kaibolayMichael G Schwern via RT wrote: > The rest of this is because, I think, FileCache does not check to see if I think you're misreading my test script. The second (and third) What I do is: { my $file = "filename"; { my $file = "filename"; [...] > FileCache does not check to see if the file has gone away and try to I don't think so. I think the "print() on unopened filehandle" is caused sub cacheout_open { Jos I. Boumans <kane@xs4all.net> replaced similar uses of caller(1) in I've updated my test script to use $file1, $file2, and $file3 as [kbolay@dose FileCacheBug]$ cat ./06export.t #For tests within the perl distribution # Functions exported by FileCache; $i = 0; # number of tests # Test 6: Test that exporting both works to package main and # First, we shouldn't be able to have these in our namespace # With an empty import list, we also shouldn't have them in ++$i; # Now, we use FileCache in 'main' ++$i; # Now we use them in another package ++$main::i; [kbolay@dose FileCacheBug]$ perl ./06export.t |
From @schwernOn Mon, Dec 13, 2004 at 04:33:17PM -0500, Kai Bolay wrote:
But they're both set to the same *filename*. Therefore, without diving too
While I'm not sure the use of caller(1) is correct, I believe that to be
Try using a different *filename*. -- |
From @schwernOn Wed, Dec 15, 2004 at 01:54:37AM -0500, Kai Bolay wrote:
Wouldn't it be easier to get it to check first if the filehandle was -- |
From @kaibolayMichael G Schwern wrote: > On Mon, Dec 13, 2004 at 04:33:17PM -0500, Kai Bolay wrote: Ah. You're right! Sorry about getting confused. Maybe FileCache can be >> I don't think so. I think the "print() on unopened filehandle" is You're right. Thanks, Kai |
|
@smpeters - Status changed from 'open' to 'resolved' |
Migrated from rt.perl.org#32949 (status was 'resolved')
Searchable as RT32949$
The text was updated successfully, but these errors were encountered: