Graham Knop [Tue, 22 Dec 2020 13:00:19 +0000]
drop Package::Stash dependency
Package::Stash refuses to work with package names that are valid, and
can break on earlier perls. Rather than working around this, just work
with the symbol tree manually. Using Package::Stash doesn't make the
code significantly cleaner or easier to understand. Additionally,
Package::Stash brings in a larger dependency tree than is reasonable for
the work it does.
Graham Knop [Tue, 30 Oct 2018 17:21:39 +0000]
clean up debugger environment before testing with it
The perl debugger uses various environment variables to configure
itself, or pick which debugger to use. Clean up those environment
variables before activating the debugger so that it behaves predictably.
Fixes RT#127516
Peter Rabbitson [Thu, 19 May 2016 19:10:46 +0000]
Adjust gitignore - we do track inc/ as of
58cd9d45
Peter Rabbitson [Wed, 18 May 2016 02:27:16 +0000]
Version 0.27
Peter Rabbitson [Tue, 17 May 2016 14:45:40 +0000]
Work around the upcoming -Dfortify_inc :/
Peter Rabbitson [Sun, 10 Apr 2016 03:04:16 +0000]
Switch to EU::HC (with an inc::latest-like approach)
Implement semi-elaborate Makefile.PL boilerplate for EU::HC use. While at
this point EU::HC seems rather stable, it still hasn't settled enough for
complete comfort.
This allow loading of "more-recent-than INC", in order to decrease amount of
needed release churn to work around newly-found problems.
Peter Rabbitson [Tue, 17 May 2016 10:30:35 +0000]
Fix silly inconsistency in docs from
253c58bc
Peter Rabbitson [Sun, 10 Apr 2016 08:55:31 +0000]
Remove incorrect SKIP for 5.8.1
The reason this test failed is srictly the construct of the test itself, and
not part of namespace::clean's functionality. reqiure() in list-assign context
combined with an @INC coderef hook would result in visible stack corruption
as per http://www.nntp.perl.org/group/perl.perl5.porters/2003/10/msg83192.html
Graham Knop [Fri, 24 Oct 2014 19:29:06 +0000]
Allow packages to be cleaned multiple times
Originally the functionality was implmented via Filter::Util::Call, but when
it was removed in [1] the "one per scope" logic remained. Given the reliance
on B::Hooks::EndOfScope none of this is longer required.
[1] https://metacpan.org/diff/file?target=PHAYLON%2Fnamespace-clean-0.07%2Flib%2Fnamespace%2Fclean.pm&source=PHAYLON%2Fnamespace-clean-0.06%2Flib%2Fnamespace%2Fclean.pm
Peter Rabbitson [Sun, 10 Apr 2016 02:09:31 +0000]
Switch to GH url in meta for metacpan's sake
See how it behaves when url/web mismatch during a trial, will fix properly
if a priblem
Graham Knop [Wed, 7 Oct 2015 22:49:34 +0000]
Restrict debugger workaround to when DB::sub will be used
The issues with the debugger only arise when DB::sub will be called.
Since the workaround requires XS modules, avoid it unless actually
needed. This avoids XS prereqs when using modules like Devel::Confess
that use debugger features but don't require the workaround.
Peter Rabbitson [Tue, 1 Mar 2016 20:13:46 +0000]
Ditch done_testing - less Test::More upgrade-churn
Peter Rabbitson [Sun, 10 Apr 2016 00:26:08 +0000]
Make sure t/10-pure-perl.t runs are identical regardless of invocation
Previously `prove -lr ./t` would trip things up
Peter Rabbitson [Sun, 11 Oct 2015 03:14:01 +0000]
Get tests running under -T (cosmetics only)
Peter Rabbitson [Fri, 9 Oct 2015 11:09:36 +0000]
Make sure $] checks run correctly (another thing my smokers didn't catch)
Peter Rabbitson [Tue, 1 Mar 2016 17:09:57 +0000]
Fix thinko from
18377f5b (RT#107813)
Peter Rabbitson [Wed, 7 Oct 2015 17:44:57 +0000]
Version 0.26
Peter Rabbitson [Wed, 7 Oct 2015 17:42:23 +0000]
Explicitly document the late runtime binding of `sort SUBNAME ...`
Peter Rabbitson [Wed, 7 Oct 2015 06:56:01 +0000]
Rewrap Changes to be more compatible with metacpan
This commit should have never existed - the way metacpan tries to parse the
changelogs is broken in the first place, on a very deep conceptual level.
Nevertheless, compromising for the sake of readability.
Peter Rabbitson [Wed, 7 Oct 2015 06:16:10 +0000]
Switch around the preference of S::U / S::N
This makes the code more complex, but will save on loading the rather heavy
B set of modules in case Sub::Util is in fact available
Peter Rabbitson [Mon, 28 Sep 2015 06:52:07 +0000]
Streamline meta etc
Peter Rabbitson [Mon, 28 Sep 2015 04:34:24 +0000]
No idea why $DebuggerFixup was originally introduced - fold it back
Also clarify Father C's original comment
Peter Rabbitson [Mon, 28 Sep 2015 06:21:16 +0000]
Make sure to exercise P::S::PP under t/10-pure-perl.t
Peter Rabbitson [Mon, 28 Sep 2015 05:59:16 +0000]
Reign in XS portions needed for certain perls under debugging, several things:
- Do not rely on Sub::Identify at all (it is too finicky) - either use
Sub::Util::subname if already loaded, or just use B (RT#96945)
- Allow either Sub::Name or Sub::Util to be used for namings
- Only add Sub::Name to the install list if neither Sub::* is available
- Clarify constant names, move them to _Util, make sure we use them everywhere
- Make sure that t/07-debugger.t runs on all perls, including the ones not
needing the XS dance
Peter Rabbitson [Sun, 27 Sep 2015 15:29:43 +0000]
Add an once-and-for-all workaround for Package::Stash::XS + =< 5.8.6
The simple workaround in
ac5c00e9 got broken by P::S::XS 0.36. Since at this
point this is more or less a pattern, add a ridiculous-ish workaround forcing
Package::Stash::PP-only on perls under 5.8.7.
While not ideal, the benefits of ::XS are virtually invisible in this use case
and moreover there is work underway to fix ::XS once and for all. Thus the
commit is considered a viable temporary measure to stave off further errors.
Peter Rabbitson [Sun, 27 Sep 2015 08:13:54 +0000]
Eject pod to the end of the file
The current file structure is very chaotic, electing to move POD out out of the
way before doing any further changes
`git blame -M` is smart enough to preserve authorship etc in such cases
Peter Rabbitson [Wed, 5 Mar 2014 11:26:15 +0000]
Version 0.25
Graham Knop [Fri, 17 Jan 2014 04:09:57 +0000]
correct EU::CBuilder check
Peter Rabbitson [Sun, 9 Dec 2012 04:30:16 +0000]
Reorganize PP test to fail comprehensibly on failed prereqs
Prompted by http://www.cpantesters.org/cpan/report/
33f54a42-3f59-11e2-b03f-503131b64f85
Peter Rabbitson [Tue, 4 Dec 2012 22:56:51 +0000]
Version 0.24
Peter Rabbitson [Fri, 23 Nov 2012 15:44:43 +0000]
B::H::EOS is now PP-capable on its own
Peter Rabbitson [Fri, 23 Nov 2012 17:26:05 +0000]
Fix tests on space-containing paths RT#77528
Peter Rabbitson [Fri, 23 Nov 2012 17:24:29 +0000]
Properly skip debugger test when optional deps not available
Peter Rabbitson [Sun, 11 Mar 2012 15:05:34 +0000]
Version 0.23
Tomas Doran [Fri, 17 Feb 2012 07:53:35 +0000]
Bump dep to B::Hooks::EndOfScope 0.10
Peter Rabbitson [Sun, 5 Feb 2012 13:20:11 +0000]
Fix failures on 5.13.6, RT#74683, Zefram++
Peter Rabbitson [Mon, 26 Dec 2011 13:01:03 +0000]
Version 0.22
Peter Rabbitson [Thu, 22 Dec 2011 23:16:24 +0000]
Whoops, more hidden packages
Peter Rabbitson [Thu, 22 Dec 2011 23:12:04 +0000]
Simplify even more - we don't care about context anymore
Peter Rabbitson [Thu, 22 Dec 2011 22:54:09 +0000]
Even more simplifications of the PP callback
Peter Rabbitson [Thu, 22 Dec 2011 11:36:37 +0000]
Another devrel to see how happy smokers are
Peter Rabbitson [Thu, 22 Dec 2011 11:20:59 +0000]
Run the PP tests with or without Devek::Hide
Peter Rabbitson [Thu, 22 Dec 2011 11:15:20 +0000]
The real fix for RT#73402 (abaondon tie %^H entirely), FC++
Peter Rabbitson [Thu, 22 Dec 2011 09:43:45 +0000]
Disable the debugger fixup entirely on perl 5.15.5+
Peter Rabbitson [Thu, 22 Dec 2011 09:29:40 +0000]
Do not rerun PP tests twice if the envvar is set
Peter Rabbitson [Wed, 21 Dec 2011 11:39:11 +0000]
Trial release to smoketest the hinthash handling changes
Peter Rabbitson [Wed, 21 Dec 2011 11:33:54 +0000]
Fix careless handling of the hinthash in PP mode (RT#73402)
Peter Rabbitson [Wed, 21 Dec 2011 10:26:20 +0000]
Add an envvar to make it easier to smoke the PP codepath
Still use Devel::Hide for the actual testing, as to not mask breakage
of the B::H::EOS detector
Peter Rabbitson [Wed, 21 Dec 2011 10:05:36 +0000]
Install the debugger fixup deps iff can_cc (RT#72368)
Peter Rabbitson [Wed, 21 Dec 2011 09:51:04 +0000]
P::S::XS buggery strikes again
Peter Rabbitson [Wed, 21 Dec 2011 09:37:45 +0000]
Fix all debugger issues from perl 5.8.1 on FC++ (RT#69862)
Peter Rabbitson [Wed, 21 Dec 2011 09:37:07 +0000]
ignore mymetas
Peter Rabbitson [Wed, 21 Dec 2011 09:07:04 +0000]
Add dropped NAME pod section (RT#70259)
Peter Rabbitson [Wed, 3 Aug 2011 21:23:20 +0000]
Version 0.21
Peter Rabbitson [Wed, 3 Aug 2011 21:46:15 +0000]
Build correct meta, not including B::H::EOS
Peter Rabbitson [Wed, 3 Aug 2011 21:04:54 +0000]
Strictures and changelogging
Peter Rabbitson [Wed, 3 Aug 2011 20:19:27 +0000]
Try harder to detect tie(%^H) clashes
Peter Rabbitson [Wed, 3 Aug 2011 19:49:48 +0000]
Clarification as per doy's request
Peter Rabbitson [Mon, 1 Aug 2011 14:02:56 +0000]
Fix borked git url
Peter Rabbitson [Sun, 31 Jul 2011 09:25:40 +0000]
Trial release preparing for pure-perl-capable 0.21
Peter Rabbitson [Sun, 31 Jul 2011 03:28:30 +0000]
De-dzilify. This may look odd and borderline revolting - read on
Currently available dzil plugins do not offer the dynconfig flexibility
that this module requires. Instead of taking a 2 day course in dzil
internals, I opted for using a tool I know to accomplish the same task.
I have absolutely no problem with someone reverting this commit and re-
doing things in dzil, as long as the resulting Makefile.PL is functionally
equivalent. Some points:
* Dzil's autorequires can't possibly understand what an optional dep is
This is an easy fix - just specify all deps manually
* There is no plugin for "require list depending on can_cc". Package::Stash
does something to the effect, but with some sort of hack I am not entirely
capable of comprehending
* There is no plugin to "require list depending on smoking env", nor there
is an easy mechanism (that I see) of combining it with the above
Peter Rabbitson [Tue, 26 Jul 2011 18:47:49 +0000]
Pure-perlize
Peter Rabbitson [Wed, 27 Jul 2011 08:55:04 +0000]
Spellfixoring RT#54388
Jason Kohles [Mon, 16 Feb 2009 22:33:25 +0000]
This used to fail with old V::M, add it to make sure it behaves under PP
Peter Rabbitson [Tue, 26 Jul 2011 13:21:02 +0000]
Moar stresstesting
Peter Rabbitson [Tue, 26 Jul 2011 14:57:49 +0000]
Only ask for debugger bits when debugger detected
Olivier Mengué [Sun, 8 May 2011 13:54:27 +0000]
Mini $RemoveSubs refactoring: less loops, less concat
Small refactoring in $RemoveSubs in the cleanup of the GLOB entry:
- grep for undefined symbols just once
- less string concatenation
Florian Ragwitz [Thu, 6 Jan 2011 18:06:46 +0000]
Version 0.20
Florian Ragwitz [Thu, 6 Jan 2011 18:04:52 +0000]
Bump Package::Stash dependency to 0.22
This pulls in a bugfix in Package::Stash::XS 0.19.
Florian Ragwitz [Mon, 3 Jan 2011 16:30:49 +0000]
Version 0.19
Florian Ragwitz [Mon, 3 Jan 2011 16:28:25 +0000]
Remove back-compat for old Package::Stash versions
0.18 is stable and doesn't have the deprecation warnings for the old API
methods. That makes it safe for us to depend on it because the upgrade won't
break existing code using it, such as Moose.
Jesse Luehrs [Sun, 2 Jan 2011 11:56:54 +0000]
work with new package::stash api
Jesse Luehrs [Sat, 13 Nov 2010 21:13:18 +0000]
don't rely on package::stash's remove_package_symbol implementation
Florian Ragwitz [Sun, 13 Jun 2010 17:10:38 +0000]
Version 0.18
Jesse Luehrs [Sun, 13 Jun 2010 16:44:14 +0000]
vivify the storage var manually
Florian Ragwitz [Sat, 5 Jun 2010 23:13:22 +0000]
Version 0.17
Florian Ragwitz [Sat, 5 Jun 2010 23:07:20 +0000]
Allow the debugger test to be skipped early
Loading the debugger often goes wrong, so it's important we skip the
test before it gets loaded.
Florian Ragwitz [Sat, 5 Jun 2010 20:52:02 +0000]
Version 0.16
Florian Ragwitz [Sat, 5 Jun 2010 20:51:34 +0000]
Stop declaring unused variables
Florian Ragwitz [Sat, 5 Jun 2010 20:49:22 +0000]
Credit doy
Florian Ragwitz [Sat, 5 Jun 2010 20:47:21 +0000]
Convert to Dist::Zilla
Jesse Luehrs [Fri, 14 May 2010 17:27:05 +0000]
changelog and version bump
Jesse Luehrs [Fri, 14 May 2010 16:52:21 +0000]
s/Stash::Manip/Package::Stash/g
Jesse Luehrs [Wed, 12 May 2010 22:47:37 +0000]
adjust deps and add changelog
Jesse Luehrs [Tue, 11 May 2010 07:48:37 +0000]
start converting to use Stash::Manip
Florian Ragwitz [Thu, 18 Mar 2010 10:17:12 +0000]
Version 0.14.
Florian Ragwitz [Thu, 18 Mar 2010 10:14:52 +0000]
Turn the very faily debugger test into an author test.
Tomas Doran (t0m) [Tue, 16 Mar 2010 17:08:07 +0000]
Remove auto_install for mbm as using git is 'too much yak hair'.
Florian Ragwitz [Sun, 17 Jan 2010 01:42:24 +0000]
Version 0.13.
Florian Ragwitz [Sun, 17 Jan 2010 01:40:26 +0000]
Skip failing debugger tests on 5.8.8 and older.
Florian Ragwitz [Thu, 14 Jan 2010 02:23:28 +0000]
Version 0.12.
Florian Ragwitz [Thu, 14 Jan 2010 02:19:44 +0000]
Remove auto-generated README.
Florian Ragwitz [Thu, 14 Jan 2010 01:51:59 +0000]
Stop relying on stash entries always being upgraded into real GVs.
Yuval Kogman [Wed, 23 Sep 2009 15:20:31 +0000]
work around $DB::sub
The variable is a string which is dereferenced as a sub that no longer
exists.
When we delete the glob where the sub was defined, we therefore alias it
to "namespace::clean::deleted::$fq" so that the debugger can still
dereference a symbolic ref and get back a working subroutine
Florian Ragwitz [Sat, 11 Jul 2009 15:31:30 +0000]
Fix restoring of non-code symbols when cleaning.
Curtesy of Ben Morrows.
Florian Ragwitz [Tue, 30 Jun 2009 05:10:50 +0000]
Add testcase for clobbering globs when trying to restore them.
Courtesy of Ben Morrows.
Florian Ragwitz [Tue, 3 Mar 2009 16:39:12 +0000]
Version 0.11.
Florian Ragwitz [Tue, 3 Mar 2009 16:32:31 +0000]
Add reference to RT ticket the recent changes fix.
Robert 'phaylon' Sedlacek [Tue, 3 Mar 2009 16:20:31 +0000]
updated README
Robert 'phaylon' Sedlacek [Tue, 3 Mar 2009 16:18:45 +0000]
removal subroutine named clean_subroutines, not remove_subroutines