Dave Rolsky [Sun, 5 Jun 2011 21:40:21 +0000]
When generating inlining code, distinguish between $default from calling $attr->default and the default val it generates
Dave Rolsky [Sun, 5 Jun 2011 21:34:04 +0000]
Add definition context to every accessor defined internally
Dave Rolsky [Sun, 5 Jun 2011 15:15:14 +0000]
ignore MYMETA.yml
Dave Rolsky [Sun, 5 Jun 2011 15:13:04 +0000]
Just fetch Value type once rather than each time inlining is done for Num and Int
Karen Etheridge [Tue, 31 May 2011 21:00:25 +0000]
update docs to show how to call into parent checks
Karen Etheridge [Tue, 31 May 2011 19:52:40 +0000]
in each subtype inline_as, use the check from the parent rather than repeating the code
Karen Etheridge [Tue, 31 May 2011 19:03:10 +0000]
Also run the special Str tests on the types that descend from Str (tests still pass).
Karen Etheridge [Wed, 1 Jun 2011 00:42:47 +0000]
fix punctuation
Karen Etheridge [Mon, 30 May 2011 20:30:12 +0000]
punctuation fix
Dave Rolsky [Thu, 19 May 2011 18:55:41 +0000]
Add Chloro to deps test
Dave Rolsky [Thu, 19 May 2011 14:49:12 +0000]
Remove JSORB
It may be what's killing Jenkins, and Stevan doesn't seem interested in maintaining it
Dave Rolsky [Wed, 18 May 2011 21:42:14 +0000]
MooseX::Net::API has been deprecated by the author
Jesse Luehrs [Thu, 19 May 2011 07:20:58 +0000]
better error message for calling superclasses as a writer when immutable
Dave Rolsky [Wed, 18 May 2011 20:36:47 +0000]
Remove FCGI::Engine
One of its tests, t/020_basic_manager.t, seem to be just hanging forever
Dave Rolsky [Wed, 18 May 2011 18:56:21 +0000]
Bump version in master
Jesse Luehrs [Wed, 18 May 2011 18:08:30 +0000]
fix changelog
Jesse Luehrs [Tue, 17 May 2011 21:33:07 +0000]
pod coverage and such
Jesse Luehrs [Tue, 26 Apr 2011 05:31:01 +0000]
add stuff from that test to author requirements
Jesse Luehrs [Tue, 26 Apr 2011 05:23:19 +0000]
tests for closing over stuff
Jesse Luehrs [Tue, 26 Apr 2011 03:04:13 +0000]
stop closing over method objects when installing modifiers
Jesse Luehrs [Tue, 26 Apr 2011 01:15:32 +0000]
also don't close over member tc objects in native delegations
Jesse Luehrs [Tue, 26 Apr 2011 00:24:49 +0000]
changelog
Jesse Luehrs [Mon, 25 Apr 2011 23:58:49 +0000]
actually, can't do that, since Moose::Error::Default isa CMOP::Object
Jesse Luehrs [Mon, 25 Apr 2011 23:37:06 +0000]
we don't need an actual inlined version of this
Jesse Luehrs [Mon, 25 Apr 2011 18:39:50 +0000]
stop closing over the type constraint object
Jesse Luehrs [Mon, 25 Apr 2011 17:48:26 +0000]
refactor the default type constraint message logic a bit
Jesse Luehrs [Mon, 25 Apr 2011 17:13:51 +0000]
stop closing over the method metaobject
changed this to close over the class metaobject if we're using a custom
error class, but there's no real way around that. the whole error system
really needs to be thrown out and redone anyway.
Jesse Luehrs [Mon, 25 Apr 2011 15:38:05 +0000]
fix setting associated_metaclass and attribute on accessor objects
Jesse Luehrs [Mon, 25 Apr 2011 07:21:14 +0000]
this only matters for lazy attributes with initializers
Jesse Luehrs [Mon, 25 Apr 2011 06:11:00 +0000]
don't close over attrs in the constructor unless necessary
Jesse Luehrs [Mon, 25 Apr 2011 05:18:48 +0000]
close over the coercion sub separately
Jesse Luehrs [Sun, 24 Apr 2011 22:47:07 +0000]
remove unnecessary(?) code
not sure what the original intent was, but it's certainly not breaking
any aliasing in this form
Jesse Luehrs [Sun, 24 Apr 2011 20:32:01 +0000]
restructure this method a bit for readability
Jesse Luehrs [Sun, 24 Apr 2011 20:14:05 +0000]
don't close over the type constraint for inlined tcs
Jesse Luehrs [Sun, 24 Apr 2011 19:39:39 +0000]
stop closing over $attr in accessor generation
Jesse Luehrs [Tue, 17 May 2011 21:04:54 +0000]
fix up a few spelling and pod issues
Dave Rolsky [Mon, 25 Apr 2011 01:57:10 +0000]
Wrap all TC inlining in parens and a do { } block
The parens protect against precedence issues when being used with logical operators.
The do { } block provides each TC its own scope, so that if it defines variables they don't leak out to the enclosing code.
Dave Rolsky [Mon, 25 Apr 2011 01:40:41 +0000]
Union inlining should always be surrounded by its own set of parens, for safety
Jesse Luehrs [Thu, 21 Apr 2011 21:35:25 +0000]
fix up the inlining test
Jesse Luehrs [Thu, 21 Apr 2011 21:32:14 +0000]
make sure these are defined
Jesse Luehrs [Thu, 21 Apr 2011 21:23:25 +0000]
these can be undef
Jesse Luehrs [Thu, 21 Apr 2011 21:16:36 +0000]
there's no reason the suffix has to be fixed length
Jesse Luehrs [Thu, 21 Apr 2011 21:15:16 +0000]
one more constrained scope in tc inlining
Dave Rolsky [Thu, 21 Apr 2011 20:55:22 +0000]
Make sure that every type passes all its tests when unioned with itself
Dave Rolsky [Thu, 21 Apr 2011 20:51:36 +0000]
Make sure that inlining union preserves the inline env
Make sure that an enum's inline env works when unioned with another enum (each env has a unique var)
Dave Rolsky [Thu, 21 Apr 2011 20:40:08 +0000]
Whenever we inline a type constraint, we need to include its inline environment.
Test every builtin TC as a simple attr's constraint, and as a native Array trait's member TC.
This thoroughly tests that we are propogating the TC's inline environment properly.
Dave Rolsky [Thu, 21 Apr 2011 20:39:20 +0000]
Make inline constraints safer ...
If we're defining a temp var - do this in a do {} block to limit its scope and
prevent it from conflicting with another constraint's var.
Don't use return in an inline constraint (this is super bad)
Don't use $_ to iterate over a collection of values in generated code, since
this could easily conflict with the surrounding code.
Dave Rolsky [Thu, 21 Apr 2011 20:33:28 +0000]
Don't assume that value that fails a duck type constraint is blessed or we get undef warnings
Dave Rolsky [Thu, 21 Apr 2011 20:18:07 +0000]
Don't use $_ when checking member constraints, since that can conflict with $_ used in inlined TC
Dave Rolsky [Thu, 21 Apr 2011 19:52:22 +0000]
Add comprehensive test for all builtin types as a simple class attribute and as a native trait attribute
Dave Rolsky [Thu, 21 Apr 2011 18:54:47 +0000]
Docs, Delta, and Changes for inline constraints
Dave Rolsky [Thu, 21 Apr 2011 18:44:24 +0000]
Rename has_inlined_type_constraint to can_be_inlined
Jesse Luehrs [Thu, 21 Apr 2011 14:13:35 +0000]
changelog
Jesse Luehrs [Thu, 21 Apr 2011 14:02:04 +0000]
one more Eval::Closure use
Jesse Luehrs [Thu, 21 Apr 2011 06:03:39 +0000]
one more fix for unoptimized constraints
Jesse Luehrs [Thu, 21 Apr 2011 05:55:51 +0000]
stop using a package global for enums, just close over it
Jesse Luehrs [Thu, 21 Apr 2011 05:55:03 +0000]
also use Eval::Closure in the test
Jesse Luehrs [Thu, 21 Apr 2011 05:40:22 +0000]
fix a couple of unoptimized constraints
there are still more that need to be fixed i think, but do we really
care about these at all? they're pretty much only useful for
informational purposes, i don't think there's any way to actually
trigger this codepath
Jesse Luehrs [Thu, 21 Apr 2011 05:39:35 +0000]
confess here, since this should never happen
Jesse Luehrs [Thu, 21 Apr 2011 05:39:23 +0000]
more readable inlined code
Jesse Luehrs [Thu, 21 Apr 2011 01:41:12 +0000]
use Eval::Closure rather than doing string eval directly
Jesse Luehrs [Thu, 21 Apr 2011 01:35:03 +0000]
compare references with == rather than eq
Dave Rolsky [Fri, 15 Apr 2011 16:11:07 +0000]
Add tests for a fake regex object ( bless {}, 'Regexp' )
Dave Rolsky [Fri, 15 Apr 2011 16:10:00 +0000]
Don't disable 'once' warnings in the whole test
Dave Rolsky [Fri, 15 Apr 2011 16:08:33 +0000]
Test that Str can handle return value of substr()
Dave Rolsky [Fri, 15 Apr 2011 16:03:20 +0000]
Make sure that Int rejects "1\n" and "\n1" (but Num does accept it)
Dave Rolsky [Fri, 15 Apr 2011 15:40:12 +0000]
A name-only subtype now inherits inlining from its parent type
Remove explicit optimized constraint in M::M::TC subclasses
Dave Rolsky [Fri, 15 Apr 2011 15:40:04 +0000]
Add inlining code for Role constraint class
Dave Rolsky [Fri, 15 Apr 2011 15:37:21 +0000]
Explicitly set the constraint for role type
Dave Rolsky [Fri, 15 Apr 2011 15:36:24 +0000]
Enum needs to check that the value is not a ref
Dave Rolsky [Fri, 15 Apr 2011 15:23:29 +0000]
Drop optimized subs for all builtins - instead use the inlining code to generate a sub
Fix inline code for ArrayRef & HashRef type - need to check the ref type of
the value before dereferencing it
Fix TODO test to test the right thing in inlining.t
Dave Rolsky [Mon, 11 Apr 2011 15:33:15 +0000]
Add a thorough test for inlining TCs, especially testing parameterized types and unions
Dave Rolsky [Mon, 11 Apr 2011 15:33:06 +0000]
Export inline_as
Dave Rolsky [Mon, 11 Apr 2011 15:32:59 +0000]
Use Moose->throw_error to throw errors
Dave Rolsky [Sun, 10 Apr 2011 23:56:00 +0000]
Use an inlined constraint when checking member values for native collection
Dave Rolsky [Sun, 10 Apr 2011 23:55:24 +0000]
We don't want two different inline code chunks to use the same var names or else they can clash and cause a warning
This happens in the case of a type union of Str | Int, for example
Dave Rolsky [Sun, 10 Apr 2011 20:42:49 +0000]
Remove extra newline
Dave Rolsky [Sun, 10 Apr 2011 20:42:25 +0000]
Add inlining for union types
Dave Rolsky [Sun, 10 Apr 2011 20:29:16 +0000]
All TC objects (except unions) now have inlining code, and tests for all the variations on calling.
Along the way, fixed the basic constraint generation for the various TC
subclasses, which were not actually setting a proper constraint sub when they
were created.
Dave Rolsky [Sun, 10 Apr 2011 16:40:18 +0000]
Inlining parameterized type seems to work, but needs more explicit tests
Dave Rolsky [Sun, 10 Apr 2011 16:20:09 +0000]
Add tests for un-parameterized Maybe
Add inlining for parameterizable types - add infrastructure for inlining parameterized, but it needs tests
Dave Rolsky [Sun, 10 Apr 2011 15:57:19 +0000]
Rewrite builtin type tests to test optimized, unoptimized, and inlined version of every constraint.
FileHandle is not subtype of GlobRef, since an IO::Handle object does not pass the GlobRef constraint.
Dave Rolsky [Sun, 10 Apr 2011 15:41:35 +0000]
Remove use of confess when we should use Moose->throw_error
Dave Rolsky [Sun, 10 Apr 2011 03:11:59 +0000]
All non-parameterized types now have inlining code
All tests pass, but we really need explicit tests for unoptimized, optimized, and inline cases
Dave Rolsky [Sun, 10 Apr 2011 02:58:47 +0000]
Fix XS for RegexpRef
Dave Rolsky [Sun, 10 Apr 2011 02:31:17 +0000]
Move optimized constraint subs to new Builtins package
Dave Rolsky [Sun, 10 Apr 2011 02:28:15 +0000]
Move definition of built in types to a separate package just for sanity
Karen Etheridge [Tue, 17 May 2011 23:32:32 +0000]
typo
Jesse Luehrs [Tue, 17 May 2011 18:14:09 +0000]
pod spelling
Jesse Luehrs [Tue, 17 May 2011 18:13:31 +0000]
edits
Jesse Luehrs [Tue, 17 May 2011 17:30:17 +0000]
update our support policy with regards to supported perl versions
Jesse Luehrs [Mon, 16 May 2011 20:13:46 +0000]
be quieter when cleaning, to make listdeps work better
Dave Rolsky [Sun, 15 May 2011 16:34:46 +0000]
Add parens around qw() to avoid warning with 5.14
Florian Ragwitz [Sun, 15 May 2011 08:16:18 +0000]
Make the changelog lie less
Florian Ragwitz [Sat, 14 May 2011 14:47:35 +0000]
Assert the correct version of CMOP being installed
Jesse Luehrs [Sat, 14 May 2011 03:30:49 +0000]
pod coverage
Jesse Luehrs [Fri, 13 May 2011 17:18:00 +0000]
changelog
Jesse Luehrs [Fri, 13 May 2011 16:34:10 +0000]
more complete clone implementation
Jesse Luehrs [Fri, 13 May 2011 16:12:40 +0000]
more tests
Florian Ragwitz [Fri, 13 May 2011 11:42:17 +0000]
Make instance cloning go through attribute setters
This way things like weakening will be taken care of.