There was an inherent race condition between edit, close & open commands which could not be prevented solely using locks. `EditFileCmd` triggered `EnsureCompiledJob` which was applying edits collected over time. At the same `CloseFileCmd` and `OpenFileCmd` were executed asynchronously and required locks on compilation unit and file lock.
Additionally, open file was resetting the module's runtime source irrespective of any edits that could already have been applied with the asynchronous execution in `EnsureCompiledJob`. This was visible especially during early manipulation of the project when open/close was performed due to a bug in IDE (#6843).
Now commands can be run either synchronously or asynchronously. Only that way can we ensure that `close` & `open` commands finish by the time any editions are being applied to module's sources.
Closes#6841.
# Important Notes
In the given video, `"foo"` would be greyed out because it would never be part of the module's (runtime) sources. Therefore no IR would be generated for it or instrumentation, meaning it would be present in `expressionUpdates` information necessary for IDE.
[Kazam_screencast_00014.webm](https://github.com/enso-org/enso/assets/292128/226a17b8-729a-415a-803f-003a9695b2f1)
close#6900
This is a follow-up to the discussion of the imports/exports meeting.
Right now we have no control over the visibility of atom constructor arguments. One way to hide them is a convention of filtering getters by an `internal` prefix or suffix.
Related to #6912
It essentially solves it by removing any builtins that would take an EnsoDate/EnsoTimeOfDay/EnsoTimeZone and replacing them with Java utils that do the same operation.
This is not a proper solution - the builtin conversion is still invalid for the date/time types - but at this moment we may just no longer use the invalid conversion so it is much less of an issue. We still need to be aware of this if we want to introduce builtins taking date/time in the future.
At the beginning of the execution `EnsureCompiledJob` acquired write compilation lock. When compiling individual modules it would then
- acquire file lock
- acquire read compilation lock
The second one was spurious since it already kept the write lock. This sequence meant however that `CloseFileCmd` or `OpenFileCmd` can lead to a deadlock when requests come in close succession. This is because commands:
- acquire file lock
- acquire read compilation lock
So `EnsureCompiledJob` might have the (write) compilation lock but the commands could have file lock. And the second required lock for either the job or the command could never be acquired.
Flipping the order did the trick.
Partially solves #6841.
# Important Notes
For some reason we don't get updates for the newly added node, as illustrated in the screenshot, but that could be related to the close/open action. Will need to dig more.
Previously, a `RuntimeException` would be thrown when an attempt would be made to curry a conversion function. That is problematic for IDE where `executionFailed` means we can't enter functions due to lack of method pointers info.
Closes#6897.
# Important Notes
A more generic solution that allows to recover from execution failures will need a follow up.
close#6800
Update the `executionContext/expressionUpdates` notification and send the list of not applied arguments in addition to the method pointer.
# Important Notes
IDE is updated to support the new API.
Throwing `TailCallException` meant that exceptions that were extracted from the expression before the call was made could not be appended. This change catches the `TailCallException`, adds warnings to it and propagates it further, thus ensuring that we don't loose the information.
Closes#6765.
# Important Notes
Removed workarounds introduced in stdlib.
Add diagnosis for unresolved symbols in `from ... import sym1, sym2, ...` statements.
- Adds a new compiler pass, `ImportSymbolAnalysis`, that checks these statements and iterates through the symbols and checks if all the symbols can be resolved.
- Works with `BindingsMap` metadata.
- Add `ImportExportTest` that creates various modules with various imports/exports and checks their generated `BindingMap`.
---------
Co-authored-by: mergify[bot] <37929162+mergify[bot]@users.noreply.github.com>
Co-authored-by: Jaroslav Tulach <jaroslav.tulach@enso.org>
The change adds an additional field to `ExpressionUpdates` messages sent by `ProgramExecutionSupport` to indicate if the type of value (or its method pointer) has changed and therefore would potentially require a suggestions' update.
Prior to #3729 that check was done during the instrumentation. However we still want to continue to support "pending expression" functionality therefore `SuggestionsHandler` will use the additional information to filter only the required expression updates.
Most of the changes are related to adapting our tests to the new field.
Closes#6706.
# Important Notes
The associated project now loads and navigates smoothly.
Also attaching a screenshot from the project that illustrates that pending functionality continues to work:
[Kazam_screencast_00006.webm](https://github.com/enso-org/enso/assets/292128/35918841-f84f-4e1c-b1b0-40e45d97e111)
close#6611
Changelog:
- update: run compiler passes on the `ascribedType` field of the constructor arguments
- update: suggestion builder uses the type information attached to `ascribedType`
- feat: resolve qualified names in type signatures
Related to #6410
# Important Notes
- Updated some `Meta` methods (needed for error handling):
- `Meta.Type` now has `name` and `qualified_name`.
- `Meta.Constructor` has `declaring_type` allowing to get the type that this constructor is associated with.
Fixes#6609 by
- e380e647af - running whole `Vector_Spec` on `java.util.ArrayList`
- 9b1229fe20 - introducing a node to handle interop values
# Important Notes
Contains additional DSL processor fix:
- 415623dcb9 - to not crash the compiler, but to properly report compiler error
Artifically limiting the number of reported warnings to 100. Also added benchmarks with random Ints to investigate perf issues when dealing with warnings (future task).
Ideally we would have a custom set-like collection that allows us internally to specify a maximal number of elements. But `EnsoHashMap` (and potentially `EnsoSet`) are still WIP when it comes to being PE-friendly.
The change also allows for checking if the limit for the number of reported warnings has been reached. It will visualize by adding an additional "Warnings limit reached." to the visualization.
The limit is configurable via `--warnings-limit` parameter to `run`.
Closes#6283.
Add format to the in-memory Column
# Important Notes
Also updates .format in date types.
Some rearrangement of date formatting builtins / Java libraries.
Engine Benchmark job runs only engine benchmarks, not Enso benchmarks.
Enso benchmarks do not report their output anywhere, and take more than 5 hours to run nowadays.
We might define a new job in the future and probably rename it to "Library benchmarks".
But that is the responsibility of the lib team.
Deeply nested modules (of depth at least 3) would have the incorrect name inferred for synthetic modules. This also became apparent in a much bigger rewrite.
Instead of having module `A.B.C`, as described in the test, it would infer the name to be `B.A.C`, which would break when trying to reference symbols from `C`.
# Important Notes
No ticket reference, as this was something that @radeusgd discovered while working on another component.
The PR includes a minimal example that would previously fail to compile.
Remove the magical code generation of `enso_project` method from codegen phase and reimplement it as a proper builtin method.
The old behavior of `enso_project` was special, and violated the language semantics (regarding the `self` argument):
- It was implicitly declared in every module, so it could be called without a self argument.
- It can be called with explicit module as self argument, e.g. `Base.enso_project`, or `Visualizations.enso_project`.
Let's avoid implicit methods on modules and let's be explicit. Let's reimplement the `enso_project` as a builtin method. To comply with the language semantics, we will have to change the signature a bit:
- `enso_project` is a static method in the `Standard.Base.Meta.Enso_Project` module.
- It takes an optional `project` argument (instead of taking it as an explicit self argument).
Having the `enso_project` defined as a (shadowed) builtin method, we will automatically have suggestions created for it.
# Important Notes
- Truffle nodes are no longer generated in codegen phase for the `enso_project` method. It is a standard builtin now.
- The minimal import to use `enso_project` is now `from Standard.Base.Meta.Enso_Project import enso_project`.
- Tested implicitly by `org.enso.compiler.ExecCompilerTest#testInvalidEnsoProjectRef`.
close#6324
Changelog
- feat: DataflowAnalysis compiler pass preserves the order of dependencies. This way when attaching the visualization to the sub-expression, the engine can find the first cached parent node, and properly invalidate it.
- update: runtime visualization test is updated to reproduce the issue
# Important Notes
The dropdown for the column `"LOCATION"` is available right after the Restaurants project startup.
Dead Letter logging is occasionally flooding our logs which is confusing to users reporting bugs. Left the possibility of a single report so that we know that something is happening.
There is 572 threads during `RuntimeServerTest` execution create. They stay around because the context isn't closed after each test finishes. Adding `afterEach` cleanup sequence to `runtime-with-instruments` tests. Improving robustness of overall `EnsoContext` [shutdown](https://github.com/enso-org/enso/pull/6468#discussion_r1180348303).
- Adjusted `Context.is_enabled` to support default argument (moved built in so can have defaults).
- Made `environment` case-insensitive.
- Bug fix for play button.
- Short hand to execute within an enabled context.
- Forbid file writing if the Output context is disabled with a `Forbidden_Operation` error.
- Add temporary file support via `File.create_temporary_file` which is deleted on exit of JVM.
- Execution Context first pass in `Text.write`.
- Added dry run warning.
- Writes to a temporary file if disabled.
- Created a `DryRunFileManager` which will create and manage the temporary files.
- Added `format` dropdown to `File.read` and `Data.read`.
- Renamed `JSON_File` to `JSON_Format` to be consistent.
(still to unit test).
Fixes#6416 by introducing `InlineableNode`. It runs fast even on GraalVM CE, fixes ([forever broken](https://github.com/enso-org/enso/pull/6442#discussion_r1178782635)) `Debug.eval` with `<|` and [removes discouraged subclassing](https://github.com/enso-org/enso/pull/6442#discussion_r1178778968) of `DirectCallNode`. Introduces `@BuiltinMethod.needsFrame` - something that was requested by #6293. Just in this PR the attribute is optional - its implicit value continues to be derived from `VirtualFrame` presence/absence in the builtin method argument list. A lot of methods had to be modified to pass the `VirtualFrame` parameter along to propagate it where needed.
During initialization JGit may attempt to resolve hostname. On some systems this can take more than desired triggering timeouts. This change does two things:
- sets the default committer for changes, lack of which probably triggered the check
- sets the default hostname to `localhost` (we don't care), in case something else in JGit still wants to resolve hostname
Closes#6447.
# Important Notes
I wasn't able to reproduce this so relying on @mwu-tow since apparently he can repro it reliably.
This change modifies method dispatch for methods that override Any's definitions. When an overrided method is invoked statically we call Any's method to stay consistent.
This change primarily addresses the plethora of problems related to `to_text` invocations. It does not attempt to completely modify method dispatch logic.
Closes#6300.
Fixes the runtime test
```
- should recompute expressions changing an execution environment *** FAILED ***
"[live]" did not equal "[design]" (RuntimeServerTest.scala:2721)
Analysis:
"[live]" -> "[design]"
```
- Missing tests from number parsing.
- Fix type signature on some warning methods.
- Fix warnings on `Standard.Database.Data.Table.parse_values`.
- Added test for `Nothing` and empty string on `use_first_row_as_names`.
- New API for `Number.format` taking a simple format string and `Locale`.
- Add ellipsis to truncated `Text.to_display_text`.
- Adjusted built-in `to_display_text` for numbers to not include type (but also to display BigInteger as value).
- Remove `Noise.Generator` interface type.
- Json: Added `to_display_text` to `JS_Object`.
- Time: Added `to_display_text` for `Date`, `Time_Of_Day`, `Date_Time`, `Duration` and `Period`.
- Text: Added `to_display_text` to `Locale`, `Case_Sensitivity`, `Encoding`, `Text_Sub_Range`, `Span`, `Utf_16_Span`.
- System: Added `to_display_text` to `File`, `File_Permissions`, `Process_Result` and `Exit_Code`.
- Network: Added `to_display_text` to `URI`, `HTTP_Status_Code` and `Header`.
- Added `to_display_text` to `Maybe`, `Regression`, `Pair`, `Range`, `Filter_Condition`.
- Added support for `to_js_object` and `to_display_text` to `Random_Number_Generator`.
- Verified all error types have `to_display_text`.
- Removed `BigInt`, `Date`, `Date_Time` and `Time_Of_Day` JS based rendering as using `to_display_text` now.
- Added support for rendering nested structures in the table viz.
related #6323
The engine can skip compilation when applying changes that do not require execution. It is more efficient to process the changes in a batch, than triggering compilations every time such edit is received.
While Enso runs single-threaded, its `ResourceManager` required additional asynchronous thread to execute its _"finalizers"_. What has been necessary back then is no longer needed since _GraalVM 21.1_. GraalVM now provides support for submitting `ThreadLocalAction` that gets then picked and executed via `TruffleSafepoint` locations. This PR uses such mechanism to _"inject"_ finalizer execution into already running Enso evaluation thread.
Requiring more than one thread has complicated Enso's co-existence with other Truffle language. For example Graal.js is strictly singlethreaded and used to refuse (simple) co-existence with Enso. By allowing Enso to perform all its actions in a single thread, the synergy with Graal.js becomes better.
`Number.nan` can be used as a key in `Map`. This PR basically implements the support for [JavaScript's Same Value Zero Equality](https://developer.mozilla.org/en-US/docs/Web/JavaScript/Equality_comparisons_and_sameness#same-value-zero_equality) so that `Number.nan` can be used as a key in `Map`.
# Important Notes
- For NaN, it holds that `Meta.is_same_object Number.nan Number.nan`, and `Number.nan != Number.nan` - inspired by JS spec.
- `Meta.is_same_object x y` implies `Any.== x y`, except for `Number.nan`.
`Vector.sort` does some custom method dispatch logic which always expected a function as `by` and `on` arguments. At the same time, `UnresolvedSymbol` is treated like a (to be resolved) `Function` and under normal circumstances there would be no difference between `_.foo` and `.foo` provided as arguments.
Rather than adding an additional phase that does some form of eta-expansion, to accomodate for this custom dispatch, this change only fixes the problem locally. We accept `Function` and `UnresolvedSymbol` and perform the resolution on the fly. Ideally, we would have a specialization on the latter but again, it would be dependent on the contents of the `Vector` so unclear if that is better.
Closes#6276,
# Important Notes
There was a suggestion to somehow modify our codegen to accomodate for this scenario but I went against it. In fact a lot of name literals have `isMethod` flag and that information is used in the passes but it should not control how (late) codegen is done. If we were to make this more generic, I would suggest maybe to add separate eta-expansion pass. But it could affect other things and could be potentially a significant change with limited potential initially, so potential future work item.
https://github.com/orgs/enso-org/discussions/6344 requested to change the order of arguments when controlling context permissions.
# Important Notes
The change brings it closer to the design doc but IMHO also a bit cumbersome to use (see changed tests) - applications involving default arguments don't play well when the last argument is not the default 🤷 .`
close#6306
Changelog:
- add: `AsyncResourceInitialization` component
- update: initialize language server resources asynchronously
# Important Notes
Speeds up `session/InitProtocolConnection` request by ~300ms (~10%) 3051ms (before) vs 2740ms (after) on my machine.
* Update type ascriptions in some operators in Any
* Add @GenerateUncached to AnyToTextNode.
Will be used in another node with @GenerateUncached.
* Add tests for "sort handles incomparable types"
* Vector.sort handles incomparable types
* Implement sort handling for different comparators
* Comparison operators in Any do not throw Type_Error
* Fix some issues in Ordering_Spec
* Remove the remaining comparison operator overrides for numbers.
* Consolidate all sorting functionality into a single builtin node.
* Fix warnings attachment in sort
* PrimitiveValuesComparator handles other types than primitives
* Fix byFunc calling
* on function can be called from the builtin
* Fix build of native image
* Update changelog
* Add VectorSortTest
* Builtin method should not throw DataflowError.
If yes, the message is discarded (a bug?)
* TypeOfNode may not return only Type
* UnresolvedSymbol is not supported as `on` argument to Vector.sort_builtin
* Fix docs
* Fix bigint spec in LessThanNode
* Small fixes
* Small fixes
* Nothings and Nans are sorted at the end of default comparator group.
But not at the whole end of the resulting vector.
* Fix checking of `by` parameter - now accepts functions with default arguments.
* Fix changelog formatting
* Fix imports in DebuggingEnsoTest
* Remove Array.sort_builtin
* Add comparison operators to micro-distribution
* Remove Array.sort_builtin
* Replace Incomparable_Values by Type_Error in some tests
* Add on_incomparable argument to Vector.sort_builtin
* Fix after merge - Array.sort delegates to Vector.sort
* Add more tests for problem_behavior on Vector.sort
* SortVectorNode throws only Incomparable_Values.
* Delete Collections helper class
* Add test for expected failure for custom incomparable values
* Cosmetics.
* Fix test expecting different comparators warning
* isNothing is checked via interop
* Remove TruffleLogger from SortVectorNode
* Small review refactorings
* Revert "Remove the remaining comparison operator overrides for numbers."
This reverts commit 0df66b1080.
* Improve bench_download.py tool's `--compare` functionality.
- Output table is sorted by benchmark labels.
- Do not fail when there are different benchmark labels in both runs.
* Wrap potential interop values with `HostValueToEnsoNode`
* Use alter function in Vector_Spec
* Update docs
* Invalid comparison throws Incomparable_Values rather than Type_Error
* Number comparison builtin methods return Nothing in case of incomparables
The primary motivation for this change was
https://github.com/enso-org/enso/issues/6248, which requested the possibility of defining `to_display_text` methods of common errors via regular method definitions. Until now one could only define them via builtins.
To be able to support that, polyglot invocation had to report `to_display_text` in the list of (invokable) members, which it didn't. Until now, it only considered fields of constructors and builtin methods. That is now fixed as indicated by the change in `Atom`.
Closes#6248.
# Important Notes
Once most of builtins have been translated to regular Enso code, it became apparent how the usage of `.` at the end of the message is not consistent and inflexible. The pure message should never follow with a dot or it makes it impossible to pretty print consistently for the purpose of error reporting. Otherwise we regularly end up with errors ending with `..` or worse. So I went medieval on the reasons for failures and removed all the dots.
The overall result is mostly the same except now we are much more consistent.
Finally, there was a bit of a good reason for using builtins as it simplified our testing.
Take for example `No_Such_Method.Error`. If we do not import `Errors.Common` module we only rely on builtin error types. The type obviously has the constructor but it **does not have** `to_display_text` in scope; the latter is no longer a builtin method but a regular method. This is not really a problem for users who will always import stdlib but our tests often don't. Hence the number of changes and sometimes lack of human-readable errors there.
close#6232
Changelog:
- remove: `SqlVersionsRepo`
- update: `SuggestionsDatabaseModuleUpdateNotification` message removing the version
- update: cleanup versions repo usages in the language server
As per design, IOContexts controlled via type signatures are going away. They are replaced by explicit `Context.if_enabled` runtime checks that will be added to particular method implementations.
`production`/`development` `IOPermissions` are replaced with `live` and `design` execution enviornment. Currently, the `live` env has a hardcoded list of allowed contexts i.e. `Input` and `Output`.
# Important Notes
As per design PR-55. Closes#6129. Closes#6131.
close#6080
Changelog
- add: implement `SuggestionsRepo.insertAll` as a batch SQL insert
- update: `search/getSuggestionsDatabase` returns empty suggestions. Currently, the method is only used at startup and returns the empty response anyway because the libs are not loaded at that point.
- update: serialize only global (defined in the module scope) suggestions during the distribution building. There's no sense in storing the local library suggestions.
- update: sqlite dependency
- remove: unused methods from `SuggestionsRepo`
- remove: Arguments table
# Important Notes
Speeds up libraries loading by ~1 second.
Implements #6134.
# Important Notes
One can define lazy atom fields as:
```haskell
type Lazy
Value ~x ~y
```
the evaluation of the `x` and `y` fields is then delayed until they are needed. The evaluation happens once. Then the computed value is kept in the atom for further use.
Fixes#5898 by removing `Catch.panic` and speeding the `sieve.enso` benchmark from 1058 ms to 514 ms. Should there be no dedicated conversion, let's use one defined on `Any` type - e.g. defining a conversion `from(Any)` makes such a conversion is always available.
Delay creation of `EnsoFile` until it is needed.
# Important Notes
By putting breakpoint into `Atom` constructor I realized few `EnsoProjectNode` instances may be created when parsing the project. It makes no sense to also create `EnsoFile` for them - until it is needed.
Modification to various tests disabled when #5917 was integrated to pass with new parser. Fixes#5894.
# Important Notes
Some tests can be fixed just by changes on the `IR` side. Some (especially error simulating ones) would benefit from changes in the `Tree` structure or at least @kazcw evaluation.
close#6139close#6137
When the project is renamed, the engine cleans up affected modules and initiates modules re-indexing to fill the suggestions database with new records. This way it reduces the amount of information stored in the suggestions database and helps implement #6080 optimization.
Changelog:
- remove: rename features from the suggestions database
- update: rename command to initiate modules cleanup and project re-execution
- fix: #6137
`--compile` command would run the compilation pipeline but silently omit any encountered errors, thus skipping the serialization. This maybe was a good idea in the past but it was problematic now that we generate indexes on build time.
This resulted in rather obscure errors (#6092) for modules that were missing their caches.
The change should significantly improve developers' experience when working on stdlib.
# Important Notes
Making compilation more resilient to sudden cache misses is a separate item to be worked on.
Avoid displaying just `Execution finished with an error: java.lang.NullPointerException` without any additional detail.
# Important Notes
When there is an execution error outside of Enso code (identified by the fact that all stack frames belong to `java` language), let's print the whole stack rather than printing nothing. To simulate one can:
```diff
diff --git engine/runtime/src/main/scala/org/enso/compiler/Compiler.scala engine/runtime/src/main/scala/org/enso/compiler/Compiler.scala
index ff0636bc66..42e5eae32e 100644
--- engine/runtime/src/main/scala/org/enso/compiler/Compiler.scala
+++ engine/runtime/src/main/scala/org/enso/compiler/Compiler.scala
@@ -453,7 +453,8 @@ class Compiler(
pool
)
} else {
- CompletableFuture.completedFuture(ensureParsedAndAnalyzed(module))
+ // CompletableFuture.completedFuture(ensureParsedAndAnalyzed(module))
+ CompletableFuture.completedFuture(())
}
}
```
It is sometimes impossible to figure out the real reason for invalid text edit request. Added a bit of context to failures to narrow down the cause of the failure.
# Important Notes
Should help with diagnosing issues like #6099.
The primary delivery of this PR is a design of `SerdeCompilerTest` - a testing suite that allows us to write sample projects, parse them with and without caches and verify they still produce the same `IR`. This is a similar idea to #3723 which compared the old and new parser `IR`s.
With infrastructure like this we can start addressing #5567 without any (significant) fear of breaking something essential.
Treat `Boolean.False` and `Boolean.True` as the corresponding primitives. Now, `Boolean.False == False` returns true.
# Important Notes
`False` and `True` constructs, that are converted to `ConstructorNode` during Truffle codegen, are handled specially in `ConstructorNode`. The easiest fix was to implement a similar special handling in `QualifiedAccessorNode`, although not the cleanest one.
A better solution would be to provide transformation of `Boolean.True` IR to a true literal in `ApplicationSaturation` compiler pass. But `ApplicationSaturation` pass does not handle `True`. Moreover, for our case, it is unnecessarily complicated.
Instrumentation of calls involving warning values never really worked because:
1) newly created nodes didn't set the UUID of their children
2) the instrumentable wrappers always had an empty (i.e. null) UUID and
they never referred `get`/`setId` calls to their delegates
On the surface, everything worked fine. Except when one actually relied on the instrumentation of values with warnings for proper setup. Then no instrumentation (replacement of nodes) was performed due to empty UUID (as required by `hasTag` of `FunctionCallInstrumentationNode`).
Closes#6045. Discovered in #5893.
