We already did this translation in the other direction, but we had no
mapping from `Optional(T)` to anything, so round-tripping a nullable
pointer from C++ through Carbon and back to C++ was previously rejected.
When importing an Interface or NamedConstraint, walk the block of
`RequireImplsId`s, and for each one:
- Import the RequireImplsDecl from it, which also imports the
`RequireImpls` structure and its id.
- Collect those decls and build a block of `RequireImplsId`s for the
local SemIR to reference from the Interface or NamedConstraint.
The import of RequireImplsDecl is done in a single phase instead of
three, unlike other decls. This is possible since require declarations
have no name, so they can't be referenced by instructions inside them,
thus there's no cycles to concern ourselves with.
Give TupleLiteral and StructLiteral a constant value, if their contents
have constant values. Their constant values are TupleValue and
StructValue respectively. This supports their ability to convert to a
constant type (or facet type).
This way when deduce finds a TupleLiteral as the argument to a
_symbolic_ facet type, it can also find a constant value to use for that
argument. This allows deduction to move onto step two, where it can
substitute into the symbolic parameter from previous deduced arguments,
and then perform the conversion from the TupleValue to the desired facet
type.
Allow `PerformBuiltinConversion()` to convert from a canonical
TupleValue or StructValue to `type` instead of only from literals. Then,
also support conversion from a symbolic binding of type TupleType or
StructType to `type`.
Add a `Core.CppCompat.NullptrT` type that C++'s `nullptr_t` maps into.
Map `nullptr` to an uninitialized constant of that type -- `nullptr`
doesn't actually have any defined bits within it, despite having the
same representation as `void*`.
The `RequireDecl` instruction points, via a `RequireImplsId` to a
`RequireImpls` structure in a `ValueStore`. That structure holds the
self-type and facet type, as well as the generic id and parent scope.
`RequireImpls` is always a generic since it only appears in an
`interface` or `constraint`, which both have a generic parameter `Self`
applied to all their members.
The `RequireDecl` instruction evaluates to itself, but drops the
decl_block_id since the instructions within the `require` declaration
are not required in the canonical value which is only used for import.
And import will want to import the `RequireImpls` structure along with
the `Interface` or `NamedConstraint` structure it is in, rather than
recreate it from the decl's instructions. This also avoids repeating all
the instructions within the `require` decl in the textual semir's
constants block.
Adding the `RequireImpls` to the `Interface` or `NamedConstraint`
structure is not yet done, so they are not available for impl lookup or
import yet.
Format the entity name into the instruction name for a FacetAccessType
of a SymbolicBinding. This means (T as type) gets formatted as
`T.as_type` instead of just as `as_type` for the non-canonical
FacetAccessType instruction. The same is already true for the canonical
SymbolicBindingType.
This requires declared FacetTypes to hold NamedConstraintIds (along with
a specific) that are named in an extend or impls requirement. We add
support to stringify and formatter to display the named constraints in
the facet type, and special case when a facet type contains a single
extend named constraint, like we did for a single extend interface.
This means that `RequireIndentifiedFacetType` can now fail, if the facet
type contains a forward-declared named constraint. Add the appropriate
diagnostics for each call to this function, and note the ones that
should change to `RequireCompleteFacetType` in the future with TODOs.
We also add tests for using facet types that can or can't be identified,
or completed, with named constraints in them.
This resolves a TODO in `expr_info.cpp` by using the inst kind rather
than the bound value to track the binding's category.
Since we're churning all the `bind_name` insts in testdata anyway, I'm
also taking this opportunity to align the inst naming with the design's
terminology, by calling these insts "bindings" (this aspect of the PR is
dependent on #6231 resolving an ambiguity in that terminology). For
consistency we'll need to rename several other insts as well (see the
TODO on `RefBinding`); I'm deferring that to a separate PR to minimize
the review load, but I think those name changes are in-scope for this
review.
Type check named constraint decls and definitions. We don't correctly
error if you put a `fn` inside them. There is no support for `require`
or `alias` yet, so there's nothing useful you can do with them yet.
We have attempted to share code between `interface` and `constraint` as
they are quite similar. First by splitting out some of
handle_interface.cpp to a separate file. Second by sharing some code
paths when you want a facet type from either one, as they both turn into
a facet type.
The SymbolicBindingType refers to the type value that will be
substituted in for the BindSymbolicName, but holds onto the EntityNameId
from the BindSymbolicName instead of (or in addition to, for now) the
instruction.
The EntityNameId will be used to look in the ScopeStack to find the
witnesses either from the BindSymbolicName instruction, or other
instructions that specify `impls` constraints against the EntityName.
This will allow us to have the `T` in `I(T)` resolve to a `.Self`
reference in the type so that we get type equality with the binding's
type: `T:! I(.Self)`.
Use the `CheckIRId` as a unique identifier for the scope of an `InstId`
- if an `InstId` is created within the scope of one `CheckIRId` it must
not be used in the scope of a different `CheckIRId`.
This is achieved without extra storage, but with false negatives for
large inputs.
When an `InstId` is created, the original index of the `Inst` is XORed
with a tag derived from the `CheckIRId` to produce the final `InstId`.
When the `InstId` is used, the expected tag is XORed with the `InstId`
to get back to the original index - if the tags don't match, the
resulting index will be corrupted, likely too large - resulting in an
out of bounds index CHECK-failure.
(the tag value is derived as such:
* take the CheckIRId
* left shift one bit (padding zero)
* left shift another bit (padding 1 - used to signify that the resulting
`InstId` has a tag combined into it)
* reverse the bits
In this way, the tag is unlikely to overlap with the index for small
test cases - making it possible to separate out the `CheckIRId` from the
index in these cases to provide more meaningful debugging/CHECK
messages, and more informative `SemIR` textual dumping that can now
include the `CheckIRId` along with the `Inst`'s index in the name of an
`inst`)
The test churn here is improved printing as tagged `InstId`s can now,
with best effort (more likely for small test cases where the `CheckIRId`
and the `Inst` index aren't at risk of overlapping from the high and low
bits), render the `CheckIRId` as part of the inst's name. Going from
`instNN` to `irMM.instNN`.
---------
Co-authored-by: Richard Smith <richard@metafoo.co.uk>
After this change, we correctly increment the offset by the next switch
case type.
Before this change, we accidentally incremented the offset by
`functions()` size instead of `cpp_overload_sets()` size and vice versa.
Also sorted the switch cases according to the order of the enum, for
consistency. This might help prevent a future similar incident.
This fix prevents crashing in the newly introduced test
`multiple_too_few_args_calls`.
This also has the side effect of showing `null name` for
`cpp_overload_set_type` and `cpp_overload_set_value`, instead of having
an arbitrary name.
Examples that demonstrate the old name is arbitrary can easily be seen
in tests like `cpp_namespace.carbon` and `decayed_param.carbon`, but
careful review would show that all old names are arbitrary, though often
luckily almost make sense.
We might want to have a proper name for these, but it's beyond the scope
of this crash fixing change.
See #6156.
Part of #5915.
This is in support of a goal of changing the blanket `destroy` impl to
use (roughly):
```
private fn CanAggregateDestroy() -> type = "type.can_aggregate_destroy";
// Handles aggregate type destruction.
impl forall [AggregateDestroyT:! CanAggregateDestroy()] AggregateDestroyT as Destroy {
fn Op[addr self: Self*]() = "type.aggregate_destroy";
}
```
That isn't done here because there's still other issues that migrating
raises. What this *does* do is add the builtin functions, and in
particular, support to `FacetTypeInfo` to make `CanAggregateDestroy`
work.
The "special requirement" approach in `FacetTypeInfo` allows us to
support restricting a blanket impl under the current approach of impls.
Maybe we'll find a cleaner approach that can work in the future, but
this fits into the current model by propagating similar to other
requirements. I'm using an enum mask because we have a number of similar
things to add (e.g. copy, move) but I'm not sure we need a full vector.
A few alternatives considered were:
- Supporting syntax more like `where .Self impls
TypeCanAggregateDestroy(.Self, SupportedInterface,
UnsupportedInterface)`. I think it'd be a little cleaner, but requires
better compile-time evaluation in order to assess the type of the call.
Right now it's expected to be a `FacetType` too early to make this work,
and I was concerned about pouring too much more time down this route.
- Providing an actual interface, in particular doing name lookup back
into `Core.` for an interface. This would've added name lookup overhead,
and the question of whether an `impl` exists.
- Generating an interface. This avoids the name lookup, but would still
raise the question of whether an `impl` should also be generated. Work
I've previously done generating interfaces for class destruction also
feels complex to both write and understand (an unfortunate issue).
- Still modeling as an `ImplsConstraint`, for example by defining a
special `InterfaceId::CanAggregateDestroy = -2` similar to what we do on
other ids. I was hesitant because of how this expands the number of
modes of `InterfaceId`, and things for consuming code to watch out for,
for what feels like a relatively niche set of use-cases that are only
interface-like.
---------
Co-authored-by: Dana Jansens <danakj@orodu.net>
As proposed in [Carbon: C++ interop for overloaded functions and
function
templates](https://docs.google.com/document/d/1KUxumZtNe3mY3TsjW2s_ZADOlAaFlrtsLKHVILtqIaM/edit?tab=t.0),
Clang is used to perform the overload resolution using C++ rules, when
an overloaded C++ set is called from Carbon. Once a function is
selected, it's converted into a Carbon function and called using the
Carbon rules including argument conversions.
A single non-templated function is treated the same way as an overload
set and the same rules apply for its call.
Template functions are not supported yet.
Demo:
a) Non-templated function calls:
```c++
// --- overloads.h
auto foo(int a, short b) -> void;
auto foo(double a) -> void;
auto foo(int a) -> void;
```
```c++
// overloads.cpp
#include "overloads.h"
#include <cstdio>
auto foo(int a, short b) -> void {
printf("hello from foo_int_short(%d, %d) \n", a, b);
}
auto foo(double a) -> void { printf("hello from foo_double(%f) \n", a); }
auto foo(int a) -> void { printf("hello from foo_int(%d) \n", a); }
```
```c++
library "Main";
import Cpp library "overloads.h";
fn Run() -> i32 {
Cpp.foo(1.1 as f64);
return 0;
}
```
```
$ clang -c overloads.cpp
$ bazel-bin/toolchain/carbon compile main.carbon
$ bazel-bin/toolchain/carbon link overloads.o main.o --output=demo
$ ./demo
hello from foo_double(1.100000)
```
b) Constructors:
```c++
// --- constructor_overloads.h
class C {
public:
C();
C(int a, int b);
};
```
```c++
// constructor_overloads.cpp
#include "constructor_overloads.h"
#include <cstdio>
C::C() { printf("hello from C() \n"); }
C::C(int a, int b) { printf("hello from C(%d, %d) \n", a, b); }
```
```c++
library "Main";
import Cpp library "constructor_overloads.h";
fn Run() -> i32 {
let c1: Cpp.C = Cpp.C.C();
let c2: Cpp.C = Cpp.C.C(1, 2);
return 0;
}
```
```
$ clang -c constructor_overloads.cpp
$ bazel-bin/toolchain/carbon compile main.carbon
$ bazel-bin/toolchain/carbon link constructor_overloads.o main.o \--output=demo
$ ./demo
hello from C()
hello from C(1, 2)
```
Follow-ups:
- `Cpp.foo({})` - proper handling of struct literals as call args.
- Fix access for overloaded sets.
- Fix tests:
- Method calls: `error: missing object argument in method call
[MissingObjectInMethodCall]` in tests.
- Fix `toolchain/check/testdata/interop/cpp/import.carbon` test.
- Fix `enums` support.
- Fix `str` -> `std::string_view` mapping.
Part of #5915
This takes the debug runtime of
`toolchain/check/testdata/interop/cpp/function/arithmetic_types_bridged.carbon`
from 4.7s down to about 4s (so 15% faster overall).
There's still lots of room to improve this test which seems to be
hitting lots of pathological behaviour, but InstNamer is 30% of the
runtime, with fingerprinting's `InstFingerprinter::GetOrCompute`
consuming 10% of cycles. We reduce its impact by using a vector of
vectors instead of a Map for the cache of fingerprints. After this
change InstNamer drops below 24% of the runtime.
Also move the instruction name when giving it to `AllocateName` since it
receives std::string by value, though this doesn't show up in the
profile for the test.
This addresses/avoids the duplicate import of vtables.
I went through a few iterations/etc along the way and left them in the
commit
history for the PR in case any of them are useful to illustrate how I
got here,
or worth revisiting.
Essentially I ended up with a circularity in importing - importing the
class
imported the vtable_decl which imported the virtual functions - and then
pending
specifics of the virtual functions needed the self specific of the
enclosing
class which wasn't ready yet.
Adding ImportRef to the vtable_decl to break the cycle caused me trouble
when
naming the vtable_decl instructions - so I tried making the functions in
the
vtable unloaded ImportRefs instead. That worked, but meant that
importing a
class still was doing O(number of vtable entries) even if the vtable
wasn't
used.
So I revisited the lazy vtable_decl - figured out how to make the naming
work
(when building the vtable_ptr, even though the vtable_decl doesn't have
to be
loaded for the vtable_ptr, I force it to be loaded anyway, to load the
vtable so
it's usable by lowering, etc). And then I could go back to the old
non-lazy
loaded vtable entries (using some loaded ImportRefs in the cases where
we needed
them/had already adopted them).
Then thinking about the VtablePtr instruction, went back/forth on
exactly what
it needed - went from VtablePtr's member being a VtableDecl InstId, to a
ClassId, then back to a VtableId as it was before this patch.
Naming the instructions has one oddity, that the VtableDecl and
VtablePtr
instructions seem to need to add the pending name for the VtableId -
despite not
using the VtableId in their own name - should the inst namer be doing
this work
for parameters of instructions rather than requiring the inst to do it
deliberately? (or am I holding it wrong in some way?)
---------
Co-authored-by: Richard Smith <richard@metafoo.co.uk>
Don't convert to f64 until we know that's the type that we actually
want. Also reimplement the conversion from RealId to FloatId to perform
an exact conversion with a real check for overflow, rather than
performing an approximate conversion via the host `double` type.
Unfortunately, LLVM doesn't expose its integer mantissa and exponent to
APFloat conversion, so we convert the RealId back to a string for now.
The LLVM conversion also detects overflow only if the literal would
round to having an out-of-range exponent, not if the literal is outside
the range of values of the type as the Carbon design expects. It's not
clear to me which rule we actually want here, so for simplicitly I'm
using the LLVM rule for now.
In preparation for adding other floating-point types beyond f64.
This makes all `.Self` references in a facet type canonically the same
(which will remain true iff they refer to the same `Self` type in the
future), removing the need to do more complex comparisons between them
using the EntityName, interface, and index. This allows the comparison
of types containing `.Self` references to be done correctly regardless
of where the `.Self` appears, as such type expressions will all be
canonically equal if they otherwise equal now, regardless of whether
they are written in the context where `.Self` could have seen different
`Self` facet types.
In order to retain access to constraints on a base `.Self` facet type,
in the case of applying `where` to an existing facet type, we:
- Give the base facet type as a `RequirementBaseFacetType` constraint so
that eval of `WhereExpr` can find and copy all the constraints off of
it.
- Introduce eager/early rewrite constraint resolution, which allows a
constraint to eagerly resolve access to earlier rewrite constraints
(`where .A = () and .B = .A` is eagerly transformed into `where .A = ()
and .B = ()`) before the full constraint resolution step. This allows
use of rewrite constraints in larger type expressions, such as `where .A
= () and .B = C(.A)` and `C` will know that the argument is `()`.
In preparation for `FloatValue` being used more generally, and not only
for literals.
---------
Co-authored-by: google-labs-jules[bot] <161369871+google-labs-jules[bot]@users.noreply.github.com>
Change impls from `<interface>.impl` to `<self>.as.<interface>.impl`,
and *member* functions to `<parent scope>.<fn>` (non-member functions
exclude their parent scope). Stop special-casing builtin functions,
given the new naming scheme.
The purpose of this is to make it clearer when a member function is
being accessed and, if so, which member function. In particular, we
often access interface `Op` functions. The builtin function
special-casing was intended to help with that, but we still have lots of
`Op` functions. This particular approach should make the interactions
clearer.
This changes up queueing of block IDs a little because, in particular,
we need to process bodies of entities only after constants finish
processing. But, it should also result in less memory usage during
processing because it means we have less on the insts stack at any given
time, since we track a block rather than all instructions contained by
the block.
This switches from the `CollectNamesInBlock` approach for entities, to
instead traversing entities as they're encountered. For example, when
traversing constants, when a type is found, the entity will have its
block queued for processing.
This leads to a change in the traversal order, which affects
disambiguation done by numeric sequencing (since that's just showing the
traversal order).
This will allow for simpler "name based on name" logic. This is
something I plan to use for:
- impls: `<type>.as.<interface>.impl`
- functions: `<entity>.<member function>`
- Note an impl may be used as the entity for a bound function.
By naming the entities as they're encountered, I'll be able to rely on
the generated names rather than recalculating them.
To assist this, I'm also differentiating between the ambiguous and
disambiguated name. Otherwise, we could end up with things like
`<function>.<disambiguator>.<call>.<other disambiguator>`, where the
repeated disambiguator may not be necessary in order to get full
disambiguation. It's also a smaller delta from the current output.
Note, changing `Name` to a class felt appropriate given its shape. I was
also noticing that parts of its API were unused, and the class helps
detect unused private members.
---------
Co-authored-by: Geoff Romer <gromer@google.com>
The goal was/is to reduce the overhead for vtables in generics - the
previous representation/prior to this patch caused a new vtable to be
created in every specific which isn't generally what we want for Carbon
generics (the whole specific/generic thing is meant to avoid creating
specific versions for things that can be a generic form parameterized by
a specific instead of manifest as a unique entity per specific)
So this moves vtables to a top level object (like functions, classes,
etc). Each dynamic class will have a vtable in this list.
Classes have a `vtable_ptr` instruction in them that points to the
vtable.
The actual generic support hasn't been implemented in this patch, as
I've been struggling with just getting this part of the migration going
& wanted to get it flushed out before adding the additional
complications.
It's possible more laziness when doing cross-file importing would be
suitable - for instance if we only need to reference the vtable from
another file, but don't need to know its individual contents, it may be
beneficial for the functions in the vtable to be import_refs (or to add
another layer of indirection - so it can be a single import_ref
all-or-nothing for the functions in the vtable).
---------
Co-authored-by: Richard Smith <richard@metafoo.co.uk>
Co-authored-by: Carbon Infra Bot <carbon-external-infra@google.com>
Add check support for `for` loops following #1885. This also adds a
basic `Optional` type to the prelude, as that's necessary to support the
new `Iterate` interface.
Depends on #5688, #5697. Those PRs aren't stacked here, but this change
will crash until they land.
---------
Co-authored-by: Jon Ross-Perkins <jperkins@google.com>
This preserves the constant values of the arguments to the thunk, which
is important if the thunk requires conversion of an `IntLiteral` to some
other type. This should become unnecessary once we have form support,
but avoiding the indirection through a thunk function seems valuable
even once that support is in place.
To support this, track whether a function is a thunk on the Function
object, and if so, what the callee of the thunk is. This information is
also included in formatted SemIR when dumping the thunk.
I've been mulling the name of this, changing it and updating comments to
try and better reflect the current semantic. "Imports" reflects how
we're currently printing this in SemIR.
This refactors `CollectNamesInBlock` to pull out the lambdas and try to
give the switch its own function. I'm hoping that, on the whole, this
makes the flow a little easier to see.
This also moves `AnyBranch` handling into the switch, instead of before
the switch.
Note, I think the helper class approach is still a little complex, but I
believe it should be negligible cost. And I couldn't think of a better
way to translate the lambdas to helper functions without adding required
arguments at the call site, which I suspected might be a source of
readability friction.
- Track the `VarPattern` instruction on the `VarStorage` instruction so
that it's available for name mangling.
- Mangle global variables based on the first binding name within their
pattern.
- Give global variables external rather than internal linkage, except if
they have no bindings whatsoever in their pattern.
- To support lowering references to bindings nested within a global var,
such as for `var (x: i32, b: i32)`, add some basic initial support for
reference constant expressions. Treat a global `var` as a reference
constant, and treat an aggregate access into a reference constant as a
reference constant.
I'm trying to make the offsetting a little easier to understand, and
also get a better `requires` structure on calls. The second is for an
attempt to refactor the `Formatter` API, but also changing the `InstId`
`derived_from` requires seems helpful for clarity on what's really
happening.
InstValueKind is really just wrapping HasTypeIdMember. Rather than
exposing this as an enum, expose it as a bool since it better reflects
what's going on.
In eval.cpp, AddImportedConstant should never be called on an untyped
instruction.
In FormatInstLhs, we can also depend on whether InstNamer has assigned a
name in order to decide whether to print an instruction. This should
avoid some divergence with CollectNamesInBlock.
We also discussed restoring InstValueKind::Untyped, but that's mainly
motivated by the formatter, and the InstNamer approach gives a more
localized implementation.
Remove calls to `InstStore::GetLocId()` to build a LocId from an InstId
now that they can be constructed directly from the InstId. Most uses of
LocId are just plumbing, so this does not affect them. However places
that want to look inside the LocId do not want to work with the InstId
form. In these places, introduce `InstStore::GetResolvedLocId()` which
converts a LocId (or an InstId as an optimization) into a LocId which is
not backed by an InstId. These locations can be printed (they have a
line and column when they are a NodeId), they can have flags added to
them (`ToImplicit`, `ToTokenOnly`), they can be converted to an
underlying ImportIRInstId, or they may be `None`.
`Dump()` is made to print a resolved location instead of printing the
InstId in the location, since (at least in my experience) the resolved
location is what is interesting in debugging, and this saves manual
`MakeInstId` steps in the debugger every time a location is of interest.
The LocId constructor from InstId is made `explicit` to add clarity to
function calls passing an `inst_id` now directly instead of calling
`context.insts().GetLocId(inst_id)`. To avoid needing to construct
`SemIR::LocId(...)` explicitly in all cases though, the diagnostics code
in Check uses `DiagnosticLocId` as its template parameter which accepts
InstId as well and does the construction of LocId from it.
Because LocId now requires an explicit construction from InstId, any
callers to `AddInst()` functions will have to explicitly convert to
LocId if they had an InstId, but not if they pass a NodeId. To make this
difference clear to callers, we `requires` that the input type can be
converted to LocId. This ensures that passing an InstId results in an
error at the callsite where the InstId is passed, instead of generating
a compiler error when trying to construct `LocIdAndInst` inside
`AddInst()`, which is less clear about what went wrong and doesn't seem
entirely intentional.
---------
Co-authored-by: Jon Ross-Perkins <jperkins@google.com>
We eliminate the `FacetAccessWitness` instruction, which would sometimes
immediately evaluate to a concrete `ImplWitness`, and sometimes remain
symbolic. This instruction is now replaced by `LookupImplWitness` in all
cases. To support the same use cases, when it is evaluated,
`LookupImplWitness` will look in the self value if it's a facet value,
and attempt to return a concrete `ImplWitness` from it before looking
for an `impl` statement.
The `LookupImplWitness` instruction's value is now canonical, even when
it evaluates to a symbolic `LookupImplWitness` instruction, by
canonicalizing the self value of the lookup query. This canonicalization
unwraps `FacetAccessType` and `FacetValue` instructions to get to an
underlying canonical facet value. However we must preserve and use the
non-canonical query while evaluating the instruction in order to look
for a concrete `ImplWitness` if the query self value was a concrete
`FacetValue`. The canonicalization ensures that symbolic witnesses
obtained from a facet value are compatible with those obtained from an
impl statement, as long as the self types originate from the same
canonical facet value though they may have been narrowed.
Member access now unconditionally does a `LookupImplWitness()`
operation, instead of only sometimes doing the lookup for a final impl
declaration.
`EvalImplLookupResult` is marked `[[nodiscard]]` so that we don't
construct it and forget to return it. This was a mistake made at one
point during the creation of this PR. And the `has_concrete_value()`
method no longer has a precondition that `has_value()` is true, since we
want to look for a concrete result only in the new use of
`EvalImplLookupResult` returned from lookup into the query self facet
value.
The TODO from `FacetAccessWitness` evaluation is addressed by ensuring
the index of the witness in the `FacetValue` comes from the required
interfaces of the `FacetValue`'s type, and that the type (a `FacetType`)
is the same facet type used in the query to construct the `FacetValue`'s
witness block. This is made possible by eliminating the
`FacetAccessWitness` indirection. The lookup into a `FacetValue` happens
while evaluating `LookupImplWitness` and it does so directly on the self
value. This gives a consistent view of the witness set and the facet
type, as they both come from the same instruction.
All of this with 400 less lines of code. :)
---------
Co-authored-by: josh11b <15258583+josh11b@users.noreply.github.com>
Co-authored-by: Richard Smith <richard@metafoo.co.uk>
Richard Smith
Boaz Brickner
Dana Jansens
Geoff RomerandRichard Smith
David BlaikieandDana Jansens
Jon Ross-Perkins
Ivana Ivanovska
