Funny enough, we thought about that but only across packages. Now, the
situation gets a little tricky because of folder structure, it's easy
to define a module "foo" in `env`, `lib` and/or `validators`. From the
compiler's perspective, they all have the same name.
For recursive structures like Tuples, the span itself isn't enough to
ensure uniqueness of elements (in particular tuples) holding elements
of the same type.
This is only a start. It compiles, but with a few TODOs left open. In particular, it doesn't currently handle constants depending on other constants or functions; nor does it hoist constants.
The playground doesn't / cannot depend on aiken-project because that becomes a gigantic pain. So instead, we try to keep essential stuff inside aiken-lang when possible.
Technically, we always need a fallback just because the way the UPLC
is going to work. The last case in the handler pattern matching is
always going to be else ...
We could optimize that away and when the validator is exhaustive, make
the last handler the fallback. Yet, it's really a micro optimization
that saves us one extra if/else. So the sake of getting things
working, we always assume that there's a fallback but, with the extra
condition that when the validator is exhaustive (i.e. there's a
handler covering all purposes), the fallback HAS TO BE the default
fallback (i.e. (_) => fail).
This allows us to gracefully format it out, and also raise an error in
case where there's an extraneous custom fallback.
When there's no type annotation in a validator handler signature, we
provide default annotation to help the type-checker. However, for
spend's datum and mint policy_id, those annotations mustn't be `Data`,
but rather Option<Data> and Bytearray.
Without that, when no annotation are provided, the compiler infer
invalid types and fails with incongruous errors.
This is a little trick which detects record access and replace them
with a simple var. The var itself is the validator handler name,
though since it contains dots, it cannot be referred to by users
explicitly. Yet fundamentally, it is semantically equivalent to just
calling the function by its name.
Note that this commit also removes the weird backdoor for allowing
importing validators in modules starting with `tests`. Allowing
validators handler to be used in importable module requires more work
and is arguably useful; so we will wait until someone complain and
reconsider the proper way to do it.
Unfortunately, as documented in:
https://github.com/IntersectMBO/cardano-ledger/issues/4571
Some Option fields in the script context certificates are going to
remain set to None, at least until the next Hard fork. There's a risk
that people permanently lock their funds if they expect deposits on
registration credentials to ever be `Some`.
So, we introduce a special type that emulate an `Option` that can only
ever be `None`. We call it `Never` and it is the first type of this
kind (i.e. with constructors indexes not starting at 0).
Without that, we may encounter weird error messages when writing
validators without an explicit `else`. Since we automatically fill it
with a `fail`; without annotation, it unifies to a generic parameter.
The existing check that would look for the body being an error term is
ill-advised as it doesn't work as soon as one adds tracing, or make
the validator a parameterized validator. Plus, it may simply trigger
the wrong behavior as one can now annotate a validator with _whatever_
and get pass the type-checker by plucking a `fail` keyword as body.
The rationale is two folds:
1. It's more consistent with how we already separate the validator
name from its module.
2. Because `_` may be present in Aiken's validator's name, it is hard
to read and I am afraid it could potentially conflict later on. So
it's better to use a separator that cannot appear in validator
names.
Some remains invalid, in particular:
- We need to handle the annotated Data case, which we still parse
correctly but do nothing about any longer.
- There's also a strange behavior with opaque type turned public?
- We now consistently desugar an expect in the last position as
`Void`. Regardless of the pattern. Desugaring to a boolean value is
deemed too confusing.
- This commit also removes the desugaring for let-binding. It's only
ever allowed for _expect_ which then behaves like a side effect.
- We also now allow tests to return either `Bool` or `Void`. A test
that returns `Void` is treated the same as a test returning `True`.
This is debatable, but I would argue that it's been sufficiently
annoying for people and such a low-hanging fruit that we ought to do
something about it.
The strategy here is simple: when we find a sequence of expression
that ends with an assignment (let or expect), we can simply desugar it
into two expressions: the assignment followed by either `Void` or a
boolean.
The latter is used when the assignment pattern is itself a boolean;
the next boolean becomes the expected value. The former, `Void`, is
used for everything else. So said differently, any assignment
implicitly _returns Void_, except for boolean which return the actual
patterned bool.
<table>
<thead><tr><th>expression</th><th>desugar into</th></tr></thead>
<tbody>
<tr>
<td>
```aiken
fn expect_bool(data: Data) -> Void {
expect _: Bool = data
}
```
</td>
<td>
```aiken
fn expect_bool(data: Data) -> Void {
expect _: Bool = data
Void
}
```
</td>
</tr>
<tr>
<td>
```aiken
fn weird_maths() -> Bool {
expect 1 == 2
}
```
</td>
<td>
```aiken
fn weird_maths() -> Bool {
expect True = 1 == 2
True
}
```
</td>
</tr>
</tbody>
</table>
This is useful when splitting module for dependencies, yet without the desire to expose internal constructors and types. This merely skips the documentation generation; but doesn't prevent the hidden module from being accessible.
The idea is pretty simple, we'll just look for lines starting with Markdown heading sections, and render them in the documentation. They appear both in the sidebar, and within the generated docs themselves in between functions. This, coupled with the order preservation of the declaration in a module should make the generated docs significantly more elegant to organize and present.
The rationale is to let them in the order they are defined, so that
library authors have some freedom in how they present information. On
top of that, we'll also now parse specifi comments as section headers
that will be inserted in the sidebar when present.
As well as fixing a couple of other issues thanks to conformance
tests. Some functions like multiply_integer or verify_ed25519_signature
have also slightly changed their costing function.
There were some odd discrepancy for `integerToByteString` on the mem
side. Either 1 or about 1000 mem units off; which I couldn't quite
figure out. Yet, it proves useful to validate builtin at large and
ensure we have a valid cost model for v3.
This covers every proposal procedures but protocol parameters, this
one is yet to be done. It spans over 30+ fields, and felt like it is a
big enough piece to tackle it on its own.
Alongside a bunch of other stuff from the coverage list. In
particular, the mint transaction contains:
- reference inputs
- multiple outputs, with assets, and type-0, type-1 and type-6
addresses.
- an output with a datum hash
- an output with an inline script
- carries an extra datum witness, preimage of the embedded hash
- mint, with 2 policies purposely ordered wrongly, with 1 and 2
assets purposely ordered wrong. One of the mint is actually a
burn (i.e. negative quantity)
This is intense, as we still want to preserve the serializer for V1 &
V2, and I've tried as much as possible to avoid polluting the
application layer with many enum types such as:
```
pub enum TxOut {
V1(TransactionOutput),
V2(TransactionOutput),
V3(TransactionOutput),
}
```
Those types make working with the script context cumbersome, and are
only truly required to provide different serialisation strategies. So
instead, we keep one top-level `TxInfo V1/V2/V3` type, and we ensure
to pass serialization strategies as type wrappers.
This way, the strategy propagates through the structure up until it's
eliminated when it reaches the relevant types.
All-in-all, this strikes a correct balance between maintainability and
repetition; and it makes it possible to define _different but mostly
identical_ encoders for the various versions.
With it, I've been able to successfully encode a V3 script context and
match it against one produced using the Haskell libraries. More to
come.
Let's consider the following case:
```
type Var =
Integer
type Vars =
List<Var>
```
This incorrectly reports an infinite cycle; due to the inability to
properly type-check `Var` which is also a dependent var of `Vars`. Yet
the real issue here being that `Integer` is an unknown type.
This commit also upgrades miette to 7.2.0, so that we can also display
a better error output when the problem is actually a cycle.
The point of those tests is to ensure that blueprints are generated
properly, irrespective of the generated code. It is annoying to
constantly get those test failing every time we introduce an
optimization or something that would slightly change the generated
UPLC.
It is impossible to serialize/deserialize a Data with a negative
constructor. So the only way this can happen is by programmatically
construct a value using builtin constr_data.
While possible, it is entirely at the responsibility of the
programmer, but not malleable from an attacker who can only provide
values as 'Data' (and thus, must be decoded like others).
Cloning a 'Term' is potentially dangerous, so we don't want this to
happen by mistake. So instead, we pass in var names and turn them into
terms when necessary.
While the ledger doesn't allow deserializing negative constr value,
they are still possible at the machine level. So, we better make sure
that we don't make assumptions regarding this.
This isn't sufficient however, as the 'assignment' helper handling
code generation doesn't perform any check when patterns are vars. This
is curious, and need to be investigated further.
Let's consider the following case:
```
type Var =
Integer
type Vars =
List<Var>
```
This incorrectly reports an infinite cycle; due to the inability to
properly type-check `Var` which is also a dependent var of `Vars`. Yet
the real issue here being that `Integer` is an unknown type.
This commit also upgrades miette to 7.2.0, so that we can also display
a better error output when the problem is actually a cycle.
The spirit here is to make it easier to discover this syntax. People
have different intuition about it and the single pipe may not be the
most obvious one.
It is however the recommended syntax, and the formatter will rewrite
any of the other to it.
The syntax is as follows:
{ "bytes" = "...", "encoding" = "<encoding>" }
The following encoding are accepted:
"utf8", "utf-8", "hex", "base16"
Note: the duplicates are only there to make it easier for people to
discover them by accident. When "hex" (resp. "base16") is specified,
the bytes string will be decoded and must be a valid hex string.
This is currently extremely limited as it only supports (UTF-8)
bytearrays and integers. We should seek to at least support hex bytes
sequences, as well as bools, lists and possibly options.
For the latter, we the rework on constant outlined in #992 is
necessary.
This is less confusing that getting an 'UnknownModule' error reporting
even a different module name than the one actually being important
('env').
Also, this commit fixes a few errors found in the type-checker
when reporting 'UnknownModule' errors. About half the time, we would
actually attached _imported modules_ instead of _importable modules_
to the error, making the neighboring suggestion quite worse (nay
useless).
We figure out dependencies by looking at 'use' definition in parsed
modules. However, in the case of environment modules, we must consider
all of them when seeing "use env". Without that, the env modules are
simply compiled in parallel and may not yet have been compiled when
they are needed as actual dependencies.
We simply provide a flag with a free-form output which acts as
the module to lookup in the 'env' folder. The strategy is to replace
the environment module name on-the-fly when a user tries to import
'env'.
If the environment isn't found, an 'UnknownModule' error is raised
(which I will slightly adjust in a following commits to something more
related to environment)
There are few important consequences to this design which may not seem
immediately obvious:
1. We parse and type-check every env modules, even if they aren't
used. This ensures that code doesn't break with a compilation error
simply because people forgot to type-check a given env.
Note that compilation could still fail because the env module
itself could provide an invalid API. So it only prevents each
modules to be independently wrong when taken in isolation.
2. Technically, this also means that one can import env modules in
other env modules by their names. I don't know if it's a good or
bad idea at this point but it doesn't really do any wrong;
dependencies and cycles are handlded all-the-same.
Using 'pallas' as a dependency brings utxo-rpc other annoying dependencies such as _tokyo_. This not only makes the overall build longer, but it also prevents it to even work when targetting wasm.
- Doesn't allow pattern-matching on G1/G2 elements and strings,
because the use cases for those is unclear and it adds complexity to
the feature.
- We still _parse_ patterns on G1/G2 elements and strings, but emit an
error in those cases.
- The syntax is the same as for bytearray literals (i.e. supports hex,
utf-8 strings or plain arrays of bytes).
There are currently two zero-arg builtins:
- mkNilData
- mkNilPairData
And while they have strictly speaking no arguments, the VM still
requires that they are called with an extra unit argument applied.
While this builtin is readily available through the Aiken syntax
`[head, ..tail]`, there's no reason to not support its builtin form
even though we may not encourage its usage. For completeness and to
avoid bad surprises, it is now supported.
Fixes#964.
The original goal for this commit was to allow casting from Data on
patterns without annotation. For example, given some custom type
'OrderDatum':
```
expect OrderDatum { requested_handle, destination, .. }: OrderDatum = datum
```
would work fine, but:
```
expect OrderDatum { requested_handle, destination, .. } = datum
```
Yet, the annotation feels unnecessary at this point because type can
be inferred from the pattern itself. So this commit allows, whenever
possible (ie when the pattern is neither a discard nor a var), to
infer the type from a pattern.
Along the way, I also found a couple of weird behaviours surrounding
this kind of assignments, in particular in combination with let. I'll
highlight those in the next PR (#979).
We've never been using those 'expected' tokens captured during
parsing, which is lame because they contain useful information!
This is much better than merely showing our infamous
"Try removing it!"
- Trace-if-false are now completely discarded in compact mode.
- Only the label (i.e. first trace argument) is preserved.
- When compiling with tracing _compact_, the first label MUST unify to
a string. This shouldn't be an issue generally speaking and would
enforce that traces follow the pattern
```
label: arg_0[, arg_1, ..., arg_n]
```
Note that what isn't obvious with these changes is that we now support
what the "emit" keyword was trying to achieve; as we compile now with
user-defined traces only, and in compact mode to only keep event
labels in the final contract; while allowing larger payloads with
verbose tracing.
Actually, this has been a bug for a long time it seems. Calling any
prelude functions using a qualified import would result in a codegen
crash. Whoopsie.
This is now fixed as shown by the regression test.
This is not fully satisfactory as it pollutes a bit the prelude. Ideally, those functions should only be visible
and usable by the underlying trace code. But for now, we'll just go with it.
This commit introduces a new feature into
the parser, typechecker, and formatter.
The work for code gen will be in the next commit.
I was able to leverage some existing infrastructure
by making using of `AssignmentPattern`. A new field
`is` was introduced into `IfBranch`. This field holds
a generic `Option<Is>` meaning a new generic has to be
introduced into `IfBranch`. When used in `UntypedExpr`,
`IfBranch` must use `AssignmentPattern`. When used in
`TypedExpr`, `IfBranch` must use `TypedPattern`.
The parser was updated such that we can support this
kind of psuedo grammar:
`if <expr:condition> [is [<pattern>: ]<annotation>]`
This can be read as, when parsing an `if` expression,
always expect an expression after the keyword `if`. And then
optionally there may be this `is` stuff, and within that you
may optionally expect a pattern followed by a colon. We will
always expect an annotation.
This first expression is still saved as the field
`condition` in `IfBranch`. If `pattern` is not there
AND `expr:condition` is `UntypedExpr::Var` we can set
the pattern to be `Pattern::Var` with the same name. From
there shadowing should allow this syntax sugar to feel
kinda magical within the `IfBranch` block that follow.
The typechecker doesn't need to be aware of the sugar
described above. The typechecker looks at `branch.is`
and if it's `Some(is)` then it'll use `infer_assignment`
for some help. Because of the way that `is` can inject
variables into the scope of the branch's block and since
it's basically just like how `expect` works minus the error
we get to re-use that helper method.
It's important to note that in the typechecker, if `is`
is `Some(_)` then we do not enforce that `condition` is
of type `Bool`. This is because the bool itself will be
whether or not the `is` itself holds true given a PlutusData
payload.
When `is` is None, we do exactly what was being done
previously so that plain `if` expressions remain unaffected
with no semantic changes.
The formatter had to be made aware of the new changes with
some simple changes that need no further explanation.
This is mainly a syntactic trick/sugar, but it's been pretty annoying
to me for a while that we can't simply pattern-match/destructure
single-variant constructors directly from the args list. A classic
example is when writing property tests:
```ak
test foo(params via both(bytearray(), int())) {
let (bytes, ix) = params
...
}
```
Now can be replaced simply with:
```
test foo((bytes, ix) via both(bytearray(), int())) {
...
}
```
If feels natural, especially coming from the JavaScript, Haskell or
Rust worlds and is mostly convenient. Behind the scene, the compiler
does nothing more than re-writing the AST as the first form, with
pre-generated arg names. Then, we fully rely on the existing
type-checking capabilities and thus, works in a seamless way as if we
were just pattern matching inline.
There's no reasons for this to be a property of only ArgName::Named to begin with. And now, with the extra indirection introduced for arg_name, it may leads to subtle issues when patterns args are used in validators.
I slightly altered the way we parse import definitions to ensure we
merge imports from the same modules (that aren't aliased) together.
This prevents an annoying warning with duplicated import lines and
makes it just more convenient overall.
As a trade-off, we can no longer interleave import definitions with
other definitions. This should be a minor setback only since the
formatter was already ensuring that all import definitions would be
grouped at the top.
---
Note that, I originally attempted to implement this in the formatter
instead of the parser. As it felt more appropriate there. However, the
formatter operates on (unmutable) borrowed definitions, which makes it
annoyingly hard to perform any AST manipulations. The `Document`
returns by the format carries a lifetime that prevents the creation of
intermediate local values.
So instead, slightly tweaking the parser felt like the right thing to
do.
While we agree on the idea of having some ways of emitting events, the
design hasn't been completely fleshed out and it is unclear whether
events should have a well-defined format independent of the framework
/ compiler and what this format should be.
So we need more time discussing and agreeing about what use case we
are actually trying to solve with that.
Irrespective of that, some cleanup was also needed on the UPLC side
anyway since the PR introduced a lot of needless duplications.
This is the best we can do for this without
rearchitecting when we rewrite backpassing to
plain ol' assignments. In this case, if we see
a var and there is no annotation (thus probably not a cast),
then it's safe to rewrite to a `let` instead of an `expect`.
This way, we don't get a warning that is **unfixable**.
We are not trying to solve every little warning edge
case with this fix. We simply just can't allow there
to be a warning that the user can't make go away through
some means. All other edge cases like pattern matching on
a single contructor type with expect warnings can be fixed
via other means.
This is crucial as some checks regarding variable usages depends on
warnings; so we may accidentally remove variables from the AST as a
consequence of backtracking for deep inferrence.
microproofs