1 ## Uncontroversial Items
3 These items are reasonably well thought out, and can go in any major release.
7 RT#59478/RT#63000 - 0+ overload causes NV conversion on == on perls before
8 5.14 - this causes comparisons to fail when the number can't fit in an NV
9 without precision loss. I'd like to fix this in a more general way (forcing
10 anyone else who might be using == on tc objects to do weird things isn't very
11 good), although it's hard to test to see what actually works.
13 ### Revise MetaRole API to reunify class/role metaroles:
20 role_attribute => [ ... ],
24 If the $meta is a class, we apply the roles to the class. If it's a role, we
25 hold onto them and apply them as part of applying the role to a class.
27 To make this all work nicely, we'll probably want to track the original role
28 where a method was defined, just like we do with attributes currently. We'll
29 also need to store method modifiers with their original role, which may mean
30 adding some sort of Moose::Meta::Role::MethodModifier class.
32 For each role-specific thing (methods, attributes, etc.) we should allow a
33 `role_attribute`, `role_method`, etc. key. The common case will be that the
34 metaroles are intended for the consuming class, but we should allow for
35 metaroles on the role's metaobjects as well.
37 ### Deprecate old-style Moose extensions
39 Moose extensions that work by calling `Moose->init_meta(metaclass =>
40 'Some::Custom::Metaclass', ...)` during their own init_meta should be
41 deprecated, so they can be removed later (this should fix the issues with
42 `init_meta` generation in Moose::Exporter, see RT51561)
44 This needs to wait until the previous fix gets in, since it will hopefully
45 eliminate the need to write custom init_meta methods entirely.
47 ### Attributes in roles need to be able to participate in role composition
49 Right now, this fails with no decent workaround:
53 has foo => (is => 'ro');
64 Role attributes really need to be able to participate in role-role combination.
65 This should also fix "with 'Role1', 'Role2'" being broken when Role1 implements
66 a method as an accessor and Role2 requires that method, but at least in that
67 case you can split it into two 'with' statements with minimal loss of
70 ### Method modifiers in roles should silently add 'requires' for them
72 This shouldn't be a functionality change, just a better error message (and
73 better introspectability). This shouldn't happen if the role already contains a
74 method by that name, so it'll depend on the previous fix going in (so "has foo
75 => (is => 'ro'); around foo => sub { }" doesn't produce a 'requires' entry).
79 There's no actual reason for this not to work, and it gets asked often enough
80 that we really should just do it at some point.
82 ### use Sub::Identify instead of doing our own thing with get_code_info
84 No idea why we stopped using Sub::Identify in the past, but there's no reason
85 not to do this. We have a bug fix in our version (the isGV_with_GP thing), so
86 this should be submitted to Sub::Identify first.
90 These are things we think are good ideas, but they need more fleshing out.
92 ### Add overloading support
94 or at least, don't break existing overloading support
96 This shouldn't treat the overloading stuff as actual methods, since that's just
97 an implementation detail, but we should provide an API for add_overload,
98 get_overload, get_overload_list, etc. In particular, this would allow
99 namespace::autoclean to not break things.
101 Also, MooseX::Role::WithOverloading should probably be cored.
103 This should probably also wait for the metarole unification fix, to avoid the
104 ::WithOverloading stuff being too insane.
106 ### Actual API for metaclass extensions
108 Right now, the only way to bundle multiple metaclass traits is via
109 Moose::Exporter. This is unhelpful if you want to apply the extension to a
110 metaclass object rather than a class you're actually writing. We should come up
111 with an API for doing this.
113 ### MooseX::NonMoose in core
115 I think all of the actual issues are solved at this point. The only issue is
116 the (necessary) implementation weirdness - it sets up multiple inheritance
117 between the non-Moose class and Moose::Object, and it installs a custom
118 constructor method at 'extends' time (although perhaps this could be solved by
119 moving some of the logic back into Moose::Object::new?). Other than that, it
120 handles everything transparently as far as I can tell.
122 ### Fix attribute and method metaclass compatibility
124 So i got this wrong when rewriting it last year - right now, metaclass compat
125 checks the default attribute and method metaclasses, which is wrong. This means
126 that if a parent class does "use MooseX::FollowPBP", then attributes declared
127 in a subclass will get PBP-style accessors, which is quite surprising.
129 On the other hand, sometimes metaclasses might need to be able to say "I'm
130 going to assume that all of my attributes at least inherit from this custom
131 class", so we might need to split it into "default specified by the user" and
132 "default specified by the metaclass" and only do compat checking on the second?
133 I'm not actually sure this is a valid use case though.
135 Something that probably should be taken into account though is attributes and
136 methods that extend existing attributes or methods from a superclass should
137 inherit the metaclass of the existing one. Also not sure if this is correct,
138 but something to think about.
140 ### Rename a bunch of the public API methods
142 Right now the public API is kind of a mess - we have things like `get_method`
143 vs `find_method_by`_name (you almost always want to use the latter), there
144 being no `has_method` equivalent that checks superclasses, `get_method_list`
145 being public but only returning method names, while `_get_local_methods` is
146 private (returning method objects), and yet neither of those looks at
147 superclasses, and basically none of this naming follows any kind of consistent
150 What we really need is a consistent and easy to remember API where the method
151 that people would think to use first is the method that they actually mean.
152 Something like renaming `find_method_by_name` to `find_method`, and `get_method` to
153 `find_local_method` or something along those lines.
155 ### Clean up metaclass constructors
157 There's a _lot_ of different conventions in here. Some things to consider:
160 * allowing new( 'name', %args ) vs ( name => 'name', %args )
161 * Method->wrap vs Method->new
163 ### Move method modifiers out to an external module
165 Class::Method::Modifiers uses a different method for doing method modifiers,
166 which I'm not sure why we aren't using in Moose right now. Optionally using
167 Class::Method::Modifiers::Fast would be even better - it uses Data::Util to
168 implement XS method modifiers, which could help things a lot.
170 ### Move type constraints out to an external module
172 There's nothing about our type constraint system that requires being tied to
173 Moose - it's conceptually an entirely separate system that Moose just happens
174 to use. Splitting it out into its own thing (that Moose could extend to add
175 things like role types) would make things conceptually a lot cleaner, and would
176 let people interested in just the type system have that.
178 ### Merge Class::MOP and Moose
180 This is a long term goal, but would allow for a lot of things to be cleaned up.
181 There's a bunch of stuff that's duplicated, and other stuff that's not
182 implemented as well as it could be (Class::MOP::Method::Wrapped should be a
185 ### Fix the error system
189 More specifically, we really want exception objects.
191 ### Moose::Util::TypeConstraints vs Moose::Meta::Type{Coercion,Constraint}
193 The Util module has _way_ too much functionality. It needs to be
194 refactored so it's a thin sugar layer on top of the meta API. As it
195 stands now, it does things like parse type names (and determine if
196 they're valid), manage the registry, and much more.
198 ### Anything with a _(meta)?class method
200 Every method that returns a class name needs to become a rw attribute
201 that can be set via the constructor.
203 ## Things to contemplate
205 These are ideas we're not sure about. Prototypes are welcome, but we may never
208 ### Does applying metaroles really need to reinitialize the metaclass?
210 Seems like the logic that's actually necessary is already contained in
211 rebless_instance, and not reinitializing means that existing attributes and
212 methods won't be blown away when metaroles are applied.
214 ### Do we want to core namespace::autoclean behavior somehow?
216 This would add Variable::Magic as a required XS dep (not a huge deal at the
217 moment, since Sub::Name is also a required XS dep, but it'd be nice for Moose
218 to be able to be pure perl again at some point in the future, and I'm not sure
219 what the relative chances of Sub::Name vs Variable::Magic making it into core
220 are). If we enabled it by default, this would also break things for people who
221 have introduced Moose into legacy-ish systems where roles are faked using
222 exporters (since those imported methods would be cleaned).
224 If we decide we want this, we may want to core it as an option first ("use
225 Moose -clean" or so), and move to making it the default later.
227 ### Should using -excludes with a role add 'requires' for excluded methods?
229 It seems to make sense, since otherwise you're violating the role's API
232 ### Moose "strict" mode
234 use Moose 'strict'; This would allow us to have all sort of expensive tests
235 which can be turned off in prod.
237 ### Moose::Philosophy.pod
239 To explain Moose from a very high level
243 We certainly have enough meta-information to make pretty complete POD docs.
247 Note that some of these are fairly old, and may not be things we actually want
250 ### `t/basics/basic_class_setup.t`
252 Imports aren't automatically cleaned. Need to think about bringing
253 namespace::autoclean functionality into core.
255 ### `t/bugs/create_anon_recursion.t`
257 Loading Moose::Meta::Class (or probably a lot of other metaclasses) before
258 loading Moose or Class::MOP causes issues (the bootstrapping gets confused).
260 ### `t/bugs/handles_foreign_class_bug.t`
262 There should be a warning when delegated methods override 'new' (and possibly
265 ### `t/bugs/role_caller.t`
267 Role methods should be cloned into classes on composition so that using
268 caller(0) in a role method uses the class's package, not the role's.
270 ### `t/cmop/metaclass_incompatibility.t`
272 If a child class is created before a parent class, metaclass compatibility
273 checks won't run on the child when the parent is created, and so the child
274 could end up with an incompatible metaclass.
276 ### `t/cmop/modify_parent_method.t`
278 Modifying parent class methods after a child class has already wrapped them
279 with a method modifier will cause the child class method to retain the original
280 method that it wrapped, not the new one it was replaced with.
282 ### `t/immutable/inline_close_over.t`
284 Initializers and custom error classes still close over metaobjects.
285 Initializers do it because the initializer has to be passed in the attribute
286 metaobject as a parameter, and custom error classes can't be automatically
289 ### `t/metaclasses/moose_exporter_trait_aliases.t`
291 Renamed imports aren't cleaned on unimport. For instance:
294 use Moose has => { -as => 'my_has' };
296 # Foo still contains my_has
298 ### `t/metaclasses/reinitialize.t`
300 Special method types can't have method metaroles applied. Applying a method
301 metarole to a class doesn't apply that role to things like constructors,
304 ### `t/roles/compose_overloading.t`
306 Overload methods aren't composed during role composition (this is detailed
307 above - Add overloading support).
309 ### `t/roles/method_modifiers.t`
311 Method modifiers in roles don't support the regex form of method selection.
313 ### `t/roles/role_compose_requires.t`
315 Accessors for attributes defined in roles don't satisfy role method
316 requirements (this is detailed above - Attributes in roles need to be able to
317 participate in role composition).
319 ### `t/todo_tests/exception_reflects_failed_constraint.t`
321 Type constraint failures should indicate which ancestor constraint failed -
322 subtype 'Foo', as 'Str', where { length < 5 } should mention Str when passed an
323 arrayref, but not when passed the string "ArrayRef".
325 ### `t/todo_tests/moose_and_threads.t`
327 On 5.8, the type constraint name parser isn't thread safe.
329 ### `t/todo_tests/replacing_super_methods.t`
331 Modifying parent class methods after a child class has already wrapped them
332 with a override will cause 'super' in the child class to call the original
333 parent class method, not the one it was overridden with.
335 ### `t/todo_tests/required_role_accessors.t`
337 Role attribute accessors don't satisfy requires from roles they consume.
339 ### `t/todo_tests/role_insertion_order.t`
341 Roles don't preserve attribute insertion_order.
343 ### `t/todo_tests/various_role_features.t`
345 * Role attribute accessors don't satisfy requires from roles they consume.
346 * Role combination should produce a conflict when one role has an actual method
347 and the other role has an accessor.
348 * Role attribute accessors should not override methods in the class the role is
350 * Role attribute accessors should be delegated when a class does
352 * Delegating to a role doesn't make $class->does('Role') true.
353 * Method modifier in a role doesn't create a method requirement.
354 * Role->meta->has_method('attr_accessor') is false.
356 ### `t/type_constraints/type_names.t`
358 Type constraint object constructors don't validate the type name provided.
364 Old todo stuff which may be totally out of date.
366 ### DDuncan's Str types
370 => where { Encode::is_utf8( $_[0] ) or $_[0] !~ m/[^0x00-0x7F]/x }
371 => optimize_as { defined($_[0]) && !ref($_[0]) };
375 => where { !Encode::is_utf8( $_[0] ) }
376 => optimize_as { defined($_[0]) && !ref($_[0]) };
380 Add support for doing it with Classes which do not have a type constraint yet
383 ### type intersections
385 Mostly just for Roles
387 ### inherited slot specs
389 'does' can be added to,.. but not changed (need type unions for this)
393 a proxied attribute is an attribute which looks like an attribute, talks like
394 an attribute, smells like an attribute,.. but if you look behind the
395 curtain,.. its over there.. in that other object
397 (... probably be a custom metaclass)
401 [13:16] mst stevan: slight problem with coerce
402 [13:16] mst I only get to declare it once
403 [13:17] mst so if I'm trying to declare it cast-style per-source-class rather than per-target-class
404 [13:17] mst I am extremely screwed
406 [13:17] stevan they are not class specific
407 [13:18] stevan they are attached to the type constraint itself
408 [13:18] * stevan ponders anon-coercion-metaobjects
409 [13:18] mst yes, that's fine
410 [13:19] mst but when I declare a class
411 [13:19] mst I want to be able to say "this class coerces to X type via <this>"
412 [13:19] stevan yeah something like that
413 [13:19] stevan oh,.. hmm
414 [13:20] stevan sort of like inflate/deflate?
415 [13:20] stevan around the accessors?
416 [13:25] * bluefeet has quit (Remote host closed the connection)
418 [13:27] mst nothing like that
419 [13:27] mst like a cast
420 [13:31] mst stevan: $obj->foo($bar); where 'foo' expects a 'Foo' object
421 [13:31] mst stevan: is effectively <Foo>$bar, right?
422 [13:32] mst stevan: I want to be able to say in package Bar
423 [13:32] mst stevan: coerce_to 'Foo' via { ... };
427 ### add support for locally scoped TC
429 This would borrow from MooseX::TypeLibrary to prefix the TC with the name
430 of the package. It would then be accesible from the outside as the fully
431 scoped name, but the local attributes would use it first. (this would need support
432 in the registry for this).
434 ### look into sugar extensions
436 Use roles as sugar layer function providers (ala MooseX::AttributeHelpers). This
437 would allow custom metaclasses to provide roles to extend the sugar syntax with.
439 (NOTE: Talk to phaylon a bit more on this)
441 ### allow a switch of some kind to optionally turn TC checking off at runtime
443 The type checks can get expensive and some people have suggested that allowing
444 the checks to be turned off would be helpful for deploying into performance
445 intensive systems. Perhaps this can actually be done as an option to make_immutable?
449 * make the errors for TCs use ->message
450 * look into localizing the messages too
451 * make ANON TCs be lazy, so they can possibly be subsituted for the real thing later
452 * make ANON TCs more introspectable
457 where { (/[a-z][a-z0-9]+/i or fail('Invalid character(s)'))
458 and (length($_) >= 5 or fail('Too short (less than 5 chars)'))
460 on_fail { MyException->throw(value => $_[0], message => $_[1]) };
462 fail() will just return false unless the call is made via $tc->check_or_fail($value);
464 * and then something like this:
466 subtype Foo => as Bar => where { ... } => scoped => -global;
467 subtype Foo => as Bar => where { ... } => scoped => -local;
471 subtype Foo => as Bar => where { ... } => in __PACKAGE__ ;
473 # or (not sure if it would be possible)
475 my $Foo = subtype Bar => where { ... };
479 [17:10] <autarch> stevan: it should do it if I pass coerce => 1 as part of the attribute definition
480 [17:12] <stevan> autarch: what I am not 100% sure of is how to tell it to deep coerce and when to not
481 [17:13] <stevan> cause a basic coerce is from A to B
482 [17:13] <autarch> hmm
483 [17:13] <stevan> which is valid for collection types too
484 [17:13] <stevan> deep coercion is what you are asking for
485 [17:13] <autarch> yeah
486 [17:13] <stevan> so perhaps we add deep_coerce => 1
487 [17:13] <stevan> which will do it
488 [17:13] <autarch> that's fine for me
491 coerce_deeply => 1 # reads better
493 ### Moose::Meta::TypeConstraint::Parameter{izable,ized}
495 The relationship between these two classes is very odd. In particular,
496 this line in Parameterized is insane:
498 foreach my $type (Moose::Util::TypeConstraints::get_all_parameterizable_types()) {
500 Why does it need to loop through all parameterizable types? Shouldn't
501 it know which parameterizable type it "came from"?