[libvariable-magic-perl] annotated tag upstream/0.55 created (now 1a6a336)
gregor herrmann
gregoa at debian.org
Wed Oct 22 18:31:32 UTC 2014
This is an automated email from the git hooks/post-receive script.
gregoa pushed a change to annotated tag upstream/0.55
in repository libvariable-magic-perl.
at 1a6a336 (tag)
tagging 18e0a2d2b05767beb8733676f14bff843d410a6d (commit)
replaces upstream/0.54
tagged by gregor herrmann
on Wed Oct 22 20:27:13 2014 +0200
- Log -----------------------------------------------------------------
Upstream version 0.55
Florian Ragwitz (1):
Add repository and bugtracker url to META.yml.
Peter Rabbitson (1):
Clarify minimum perl 5.10.0 requirement for hash magic
Shlomi Fish (1):
A few doc corrections
Vincent Pit (431):
Importing Variable-Magic-0.01
Importing Variable-Magic-0.02.tar.gz
Importing Variable-Magic-0.03.tar.gz
Importing Variable-Magic-0.04.tar.gz
Importing Variable-Magic-0.05.tar.gz
Importing Variable-Magic-0.06.tar.gz
Importing Variable-Magic-0.07_01.tar.gz
Importing Variable-Magic-0.07_02.tar.gz
Importing Variable-Magic-0.08.tar.gz
Importing Variable-Magic-0.09.tar.gz
Importing Variable-Magic-0.10.tar.gz
Importing Variable-Magic-0.11.tar.gz
Importing Variable-Magic-0.12.tar.gz
Importing Variable-Magic-0.13.tar.gz
Importing Variable-Magic-0.14.tar.gz
Importing Variable-Magic-0.15.tar.gz
Importing Variable-Magic-0.16.tar.gz
Importing Variable-Magic-0.17.tar.gz
Enable uvar magic when patchlevel >= 28419
Remove old tests that were wrongly imported
Set VMG_COMPAT_ARRAY_PUSH_NOLEN for Windows ActiveState Perl 5.8.x
This is 0.18
Update .gitignore
Add the VMG_HAS_PERL_{AND,OR} macros to test for a patchlevel and/or a perl version
Replace old version tests by VMG_HAS_PERL_OR
Enforce the API change for copy callbacks only in 5.11.x
'uvar' magic will never be backported to 5.8, enable it only in maint-5.10 and higher
Remove PERL_API_* macros
Remove yet another wrongly imported tests
Export the patchlevel with the new VMG_PERL_PATCHLEVEL constant
Use the new VMG_PERL_PATCHLEVEL constant to print the patchlevel in t/00-load.t
Force temps cleanup after removing the wizard from the context hash, but flag it as SVf_BREAK so that it won't be freed again later
This is 0.19
Replace XPUSHs by EXTEND/PUSHs
Get rid of Debian_CPANTS.txt
Test caller inside callbacks
Fix the caller-into-block test that was actually testing nothing. Also use __LINE__ instead of hardcoding line numbers
Only enable turn the MGf_* flags on when the corresponding callback has been specified
Test how arguments are passed to the data callback
POD beautifications
Correct backtrace when wizard() croaks, and test it
Make the module threadsafe by adding a CLONE method that clones the global state. Test it with t/40-threads.t. Export the thread-safety status through VMG_THREADSAFE
Turn READONLY on onto wizard objects
Skip wizard destructor during global destruction
Don't test threads if this perl wasn't built with ithreads enabled
Thread detection macros overhaul
Remove refcount debugging macro
Use 0 (store as-is) as the len when casting magic on the wizard
Clone wizards across threads
Test same signature across threads
Allocate context hash iterators as soon as they are created
Test destructors
Factor signature/wizard checks into proper functions. 'cast , undef' now throws an 'invalid wizard' error
Move vmg_clone() upwards
Test cloning
Really don't try to read the context before we're sure we are not dirty
t/41-clone.t needs to call wizard into an eval block
This is 0.20
Enable thread-safety for perls that just lacks MY_CXT_CLONE
Shrink the context data by getting rid of the mostly useless .count member
Silence three 'mixed declaration and code' warnings
Remove two unused variables
Add some parenthesis to appease the compiler
Output threads and threads::shared versions in threads tests
This is 0.21_01
Hide previously deleted magic tokens in vmg_svt_free (Yuval Kogman)
This is 0.21_02
Disable thread safety for 5.8.x on Windows
A little paragraph on how magic is cool
This is 0.22
Better be on irc.perl.org
Only build clone functions when VMG_THREADSAFE is set
w->cb_dup is currently always NULL
This is 0.23
Really fix building on perls with multiplicity but no threads
This is 0.24
Old Pod::Coverage don't ignore CLONE
Version macro overhaul. Enable correct behaviour for 5.8.9
Only enable VMG_COMPAT_ARRAY_PUSH_NOLEN for ActiveState's 5.8.[78] on Windows, as the relevant patch has been integrated for 5.8.9
Improve logic for the SvREFCNT_inc workaround in vmg_data_new(). Also, enable it up to 5.8.3 included, as it was before the previous macro overhaul
This is 0.25
Fix building for maint-5.8
Document magic changes in 5.8.9
Introduce VMG_COMPAT_ARRAY_UNSHIFT_NOLEN_VOID to cover unshift no longer calling len magic in void context (since 34908)
Fix leaks of cloned coderefs that access lexicals
This is 0.26
The context is actually used in vmg_sv_magicuvar() for sv_magic()
Don't redefine VMG_COMPAT_ARRAY_PUSH_NOLEN
Fix an offset of 1 with len magic on scalars
More linefeeds make the POD clearer
Document the return values of the callbacks
Factor some test logic in a helper module
Convert t/20-get.t to the new testing framework
Convert t/21-set.t to the new testing framework
Cleanup t/22-len.t
Convert t/23-clear.t to the new testing framework
Convert t/24-free.t to the new testing framework
Convert t/25-copy.t to the new testing framework
Convert t/27-local.t to the new testing framework
Convert t/28-uvar.t to the new testing framework
Convert t/30-scalar.t to the new testing framework
Convert t/31-array.t to the new testing framework
Convert t/32-hash.t to the new testing framework, but don't check clear/copy magic interaction yet
Fix segfaults that seems to happen when one croak in a callback at compile time
Convert t/33-code.t to the new testing framework
Convert t/34-glob.t to the new testing framework
Convert t/16-huf.t to the new testing framework
Document the "get/clear on hashes" problem
Bump copyright year
This is 0.27
Test the "die in free callback in BEGIN" issue in a new t/17-ctl.t
Remove an unused variable
A better fix for handling exceptions at compile time.
Spacing
Output the build number for Win32 AS perls at the beginning of the tests
Disable the Win32 AS Perl workaround for now, as it doesn't seem to be relevant for the official PPM repository
Don't call the free callback in global destruction
Fix refcount of scalars with free callback
Report file/line of failures correctly
Be a little more precise about the get/clear problem
Test emptying a hash (can't believe this wasn't already tested)
Reenable copy tests in t/32-hash.t as they are not the issue
Also test exists in t/32-hash.t
Test assigning to the key in an uvar callback
Make the benchmark a little more fair
Allow editing the key SV in uvar callbacks by passing a new option 'copy_key'
Better documentation for the values passed to the callbacks
Show copy_key in wizard()'s synopsys
Don't assign results in check blocks
More tests for t/31-array.t
Fix segfaults when using get or uvar magic simultaneously with clear magic
Test uvar magic on tied hashes
This is 0.28
A more reliable logic for turning off the GMAGICAL flag when no uvar magic is involved
Minor POD fix
count isn't used anymore in vmg_svt_len()
Comment revamp in vmg_svt_free()
sv_mortalcopy() is more efficient than sv_2mortal(newSVsv())
Don't even define a CLONE sub when the module isn't thread safe
Test description nit in t/28-uvar.t
Amend last change to make t/28-uvar.t pass again
Set $_[2] to the default length for len magic on scalars
Use recent ExtUtils::MakeMaker features to generate a better META.yml
Require perl 5.8
This is 0.29
mPUSHi is no longer needed, but mPUSHu does
Get current op information in magic callbacks
Add compatibility macro for SvPV_const
Really skip free callbacks during global destruction
Use VMG_OP_INFO_* macros in the synopsys for op_info
Note the op argument position in wizard() synopsys
Test magic on array/hash elements
Carp is also a build dependency
Test magic on symbol table in a new t/35-stash.t
free magic isn't called for array values delete on 5.8.5 and lower
Prefer using VMG_OP_INFO_* constants in t/35-stash.t
Fix the testsuite for 5.8.3 and lower
Document magic on array/hash values
This is 0.30
Reorder a bit the wizard duplication logic to silence a cast warning
getdata() handles the stack itself, so make it return void and use PPCODE
croak() if storing the wizard in the context fails, but don't let that mess up the coverage
Magic variables can't be distinguished by tied()
Reference count vmg_op_name_len
Make op_info thread safe
A clearer introduction to magic
More code tests
'assignation' is really 'assignment'
Reorder the POD so that the functions come before the constants and the history
Stop citing all the constants in the EXPORT section
Fix and test segfaults and wrong "Unknown error" exceptions when dieing in require cause a free callback to fire
Factor the callback retval retrieval logic into a new VMG_CB_CALL_SET_RET() macro
Inline the 'eval' specific part of vmg_cb_call() into vmg_svt_free()
Explicitely pass the flags to vmg_cb_call(). Also pass data in the va_args
This is 0.31
Bless the op info object into the right class (stealing the logic from B)
More tests for different op types
An example on how to cast magic recursively
Pass the signature as a thread callback argument in t/40-threads.t
A refreshed synopsis
Move the op_info tests to a new t/18-opinfo.t
Test op_info in free callbacks
Hide an uncoverable line
A note on global destruction and free callbacks
Reorder the arguments to the thread callback in t/40-threads.t so that they match those of t/41-clone.t
Fix the B stashes cache cloning, and really use it for blessing op objects
Minor fixes in the global destruction comment
L<threads::shared> variables
Don't store the sig in the mg_private field
Test numeric signatures too big/small
Check the number of allocated wizards in vmg_gensig()
hv_exists() is enough for vmg_wizard_sig()
We don't need a (wrong) compatibility macro for NewOp, since we only use it on 5.10
Initialize the global op names cache in a critical section
clear magic only applies to arrays and hashes
Remove some trailing spaces
This is 0.32
Fix samples/synopsis.pl permissions
Store a placeholder in the wizards set for every manually generated signature.
The current op is always non-null in our vmg_opclass()
Hide the 'too many sigs' test from coverage
Test loop ops
Build better testcases in t/18-opinfo.t
Test L:{last L} in t/18-opinfo.t
Clean up the remaining if (...) { ... } tests
Factor the regexp in t/12-sig.t
Make getsig() croak if the supplied wizard is incorrect
Document the changes regarding how cast/getdata/dispell handle errors
free callbacks fire also at the end of the global scope
Lock shared variables
Revert "clear magic only applies to arrays and hashes"
Rename test functions init() and check() to init_watcher() and watch()
Test when magic actions take place
Test variables in other stashes
There's no need to disable strict refs for getting the stash of a package whose name is known at compile time
More stash tests for functions/methods
Better metasyntactic variable names for t/35-stash.t
Test stash magic with AUTOLOAD
Test that the uvar/clear doesn't confuse B::Deparse in t/32-hash.t
skip $desc => $num is neater
This is 0.33
Properly unmortalize the wizard after removing it from the global hash
Simplify the current temp test in the unmortalizer
Test (lack of) stash magic for dynamic method calls
Point the repository to the gitweb instead of the bare repo
Rename Variable::Magic::TestDieRequired to Variable::Magic::TestScopeEnd
Prevent call_sv() in vmg_svt_free() to clobber the last popped context slot
This is 0.34
Slashes are reserved and hence must be encoded in the search part of an URL
Remove the fake context namesv workaround introduced in fbcb2462798d2fa931a5c97ebf0cec73177dce23, as 135abb4d38a3a9eab3a7e6d6cde33aafd62488c2 also fixes the issue in a better way
Store the last popped stack index for after call_sv()
Test repeated method calls on stashes
Drop support for 5.9.5
Require threads 1.67 and threads::shared 1.14 in t/40-threads.t
This is 0.35
VMG_COMPAT_ARRAY_PUSH_NOLEN needs to be enabled for AS Perl 5.8 with build >= 822
Also require threads 1.67 and threads::shared 1.14 in t/41-clone.t
Make getdata() return an empty list when no magic is present
This is 0.36
Righteously is really rightfully
Clarify when get/set magic triggers for container types
Get rid of t/90-boilerplate.t
base.pm is also a dependency
Introduce VMG_FORKSAFE
This is 0.37
Kwalitee test overhaul
Give an explicit value to dynamic_config
Put prerequisites in their own separate hash
Build the $name and the main $file from $dist
Re-indent WriteMakefile() arguments
A better t/01-import.t
Introduce VMG_COMPAT_ARRAY_PUSH_NOLEN_VOID and fix 5.11.0 compatibility
Deprecate all the user signature stuff for december 2009
Add a cookbook section
This is 0.38
Compatibility fix for 5.11.1
Remove all signature-related features
Factor the V::M magic searching logic into a new vmg_find()
Remove unused variables
Empty the global pointer table at the end of a thread
Make the wizard store the MGWIZ address as an IV
Fix a consting error
Stop leaking the SVs that reference the callbacks
Free the thread local context from an atexit callback
Directly store the CVs into the MGWIZ structure
Fix t/35-stash.t for 5.11.2
Stop passing arguments to the data callback through an AV
Stop leaking objects stored in the data slot
Also test getdata() in t/80-leaks.t
Test destruction of magic callbacks
Test string callbacks
Fix compiler warnings with the Intel CC
Document passing a string reference as the callback
This is 0.39
Fix memory miswrite when passing data arguments to cast()
Wizard ids are IV, not UV
Explicitely cap the op_info parameter to 255
call_sv() ought to take an I32
Fix type coercions on values returned from callbacks
Fix expected constness for utf8_length() arguments
Use size_t as the index type in the pointer table
Cast the op_info value before applying bit shifts
SvPV_const is only available since 5.9.3
Use SvRV_const() whenever possible
Use SvREFCNT_inc_simple_void() whenever possible
Bump copyright year
Indentation
This is 0.40
Correctly propagate the errors thrown when variable destruction happens at compile-time
Tune the userdata example
Compatibility fix for 5.12.0
This is 0.41
Improve op_info coverage
Complete coverage of the len callback
Fix t/18-opinfo.t failures with perl 5.13
Fix tests for perl 5.13
This is 0.42
Exception propagation fixes
Always use a safe version of call_sv()
Fix CLONE_PARAMS struct initialization for perl 5.13.2
5.13.2 calls get magic on globs
Prettify the constants export list
More control tests
This is 0.43
SvREFCNT_dec already checks if the SV is non null
Clarify some logic in Makefile.PL
Force linking against the perl dll when using gcc 3.4 on Windows
Remove the test for non-released perl 5.11.0
Remove the patchlevel test
Skip threads tests unless perl version is 5.13.4 or greater
Force threads tests with the PERL_VARIABLE_MAGIC_TEST_THREADS environment variable
Describe why we don't have to guard against PL_dirty in vmg_mgwiz_free()
Only load Carp.pm when throwing an error
Correctly propagate exceptions from _wizard()
This is 0.44
Ignore MYMETA.yml
Explicitely state that C++ compilers aren't supported
Add ptable_delete to ptable.h
Skip NULL values when walking a table
Give a saner default to pPTBLMS
Wording fix
Add more glob tests
A bare glob that isn't part of a fetch doesn't call get magic anymore in 5.13.7
Only use Capture::Tiny when it is working on a simple example
A terser way to display versions for modules used in tests
This is 0.45
Tighten the scope of two variables unused on 5.8
Bump copyright year
Update magical flags after dispelling magic
Switch to qw<>
This is 0.46
Update vmg_opclass() for perl 5.15
Ignore MYMETA.json
Replace $] by "$]"
Encode the README file in UTF-8
Add META.json
Tweak some optional module loading in tests
Silence an "unused" compiler warning
Preserve the PATH environment variable when running a sub-perl on cygwin
Make sure code examples fit in a 80 columns terminal
Remove trailing whitespace
Threads tests may not be able to spawn all the threads
Pass the 'data' callback after the 'op_info' flag when calling _wizard()
Abstract the MGWIZ creation into a new vmg_mgwiz_alloc() function
MGf_COPY and MGf_DUP should always be set for the perls we consider
Wrap MUTEX_{LOCK,UNLOCK} in macros so can we don't have to #ifdef them
Share the vtables with threaded perls
Clean up threads tests a bit
Clone wizard objects through dup magic
Rename vmg_wizard_* to vmg_wizard_sv_* and {MGWIZ,mgwiz}_* to vmg_wizard_*
Reorder some code
Enforce saner checks on wizard SVs and MAGIC tokens
The magic signatures are no longer needed
Make the watch { } wrapper properly apply context
vmg_cast() should set mg_flags according to the vtable
Prettify the declaration of vmg_svt_copy()
Minimize the accesses to the optree during the uvar/clear hack
Clean up vmg_uvar_val() a bit
'store' and 'delete' uvar magics don't need the uvar/clear hack
Test 'delete' uvar magic and introduce VMG_COMPAT_HASH_DELETE_NOUVAR_VOID
This is 0.47
Test that unqualified sub names as callbacks default to the current package
Protect $@ inside Variable::Magic::TestWatcher::init_watcher
Move the sv length logic into a new vmg_sv_len() function
In t/22-len.t, localize the wizard for strings to the SKIP block
Allow passing ref-to-undef as callbacks to install a noop callback
Bump copyright year
Remove an useless #define wrap
Decorate the magic callbacks section
This is 0.48
Require Module::ExtractUse 0.24 for the Kwalitee test
Reuse the diag() helper in the TestThreads module
Teach t/17-ctl.t about perl 5.17.0
This is 0.49
Teach t/35-stash.t about perl 5.17.1
Jumbo POD overhaul
Remove one extra mention to MGf_COPY
Cleaner version numbers
Don't leak vtables and wizards freed during global destruction
Lower t/17-ctl.t's last test expectations
Move the global destruction logic of t/15-self.t out in an helper module
Leak less memory when an exception is thrown from a free callback
Remove a couple of commented asserts
Consistently format magic types with I<>
Don't use the main:: stash in the last test of t/17-ctl.t
Improve Capture::Tiny discovery in t/17-ctl.t
This is 0.50
Make the uvar user data into a proper struct
Always enable VMG_SAVE_LAST_CX
Port the svt_free-specific part of vmg_call_sv() to a customizable hook
Fix indentation in vmg_call_sv()
Make sure all entries of MY_CXT.b__op_stashes are initialized
Rename vmg_uvar_del() to vmg_mg_del() to reuse its logic
Don't crash if dispell is called in a clear, free or uvar callback
Prettify the default no-op callback
Remove an unused variable
Reset the SV each time for the 'reset RMG flag' workaround
Make the 'reset RMG flag' workaround thread-safe
Kill Debian_CPANTS.txt suffixes as well
Properly propagate exceptions when a free callback dies at the end of eval
Documentation tweaks
This is 0.51
Update how author tests load their dependencies
Check destruction order for exceptions thrown in free callbacks
Add BUILD_REQUIRES to WriteMakefile()
Update author tests requirements
Don't run the POD spelling test in taint mode
Simplify Config loading in Makefile.PL
Make t/35-stash.t pass with perl 5.17.4 and above
Add VMG_COMPAT_SCALAR_NOLEN
This is 0.52
Update VPIT::TestHelpers to e8344578
Typo in POD
Update the bug tracker URL in META after the rt.perl.org upgrade
Get rid of t/99-kwalitee.t
Stop bundling author tests
No tabs please
Remove obsolete kwalitee workarounds
Make Perl version numbers more readable
Bump copyright year
This is 0.53
Make t/35-stash.t pass on 5.21.4
Use run_perl() from VPIT::TestHelpers
Really make t/35-stash.t pass on 5.21.4
Add support for copy magic on code prototype clone
Bump copyright year
This is 0.54
Don't hardcode the OPc_* values
Add support for METHOP ops
Remove test dependencies list in POD
This is 0.55
gregor herrmann (1):
Imported Upstream version 0.55
-----------------------------------------------------------------------
No new revisions were added by this update.
--
Alioth's /usr/local/bin/git-commit-notice on /srv/git.debian.org/git/pkg-perl/packages/libvariable-magic-perl.git
More information about the Pkg-perl-cvs-commits
mailing list