The main culprit of a Vector slowdown (when compared to Array) was the normalization of the index when accessing the elements. Turns out that the Graal was very persistent on **not** inlining that particular fragment and that was degrading the results in benchmarks.
Being unable to force it to do it (looks like a combination of thunk execution and another layer of indirection) we resorted to just moving the normalization to the builtin method. That makes Array and Vector perform roughly the same.
Moved all handling of invalid index into the builtin as well, simplifying the Enso implementation. This also meant that `Vector.unsafe_at` is now obsolete.
Additionally, added support for negative indices in Array, to behave in the same way as for Vector.
# Important Notes
Note that this workaround only addresses this particular perf issue. I'm pretty sure we will have more of such scenarios.
Before the change `averageOverVector` benchmark averaged around `0.033 ms/op` now it does consistently `0.016 ms/op`, similarly to `averageOverArray`.
Trying to invoke a foreign method with non-installed language (either not enabled in the Truffle `Context`, or not installed in the GraalVM distribution) results in `Polyglot_Error`, rather than crashing the entire engine.
- Reimplement the `Duration` type to a built-in type.
- `Duration` is an interop type.
- Allow Enso method dispatch on `Duration` interop coming from different languages.
# Important Notes
- The older `Duration` type should now be split into new `Duration` builtin type and a `Period` type.
- This PR does not implement `Period` type, so all the `Period`-related functionality is currently not working, e.g., `Date - Period`.
- This PR removes `Integer.milliseconds`, `Integer.seconds`, ..., `Integer.years` extension methods.
This PR introduced an overhauled Component List Panel implementation, making use of the efficient EnsoGL grid view component. Also, it delivers a couple of new features:
* A part of the new design: there are no more section headers in grid, instead groups are "glued" together. The local scope section is under "popular" (old "favorites").
* The keyboard management inside grid works.
* there is a mouse hover highlight
* selecting the lowest entry in section when jumping with navigation bar.
* accepting input as-is with cmd/ctrl + Enter.
https://user-images.githubusercontent.com/3919101/194561890-fffb9b41-2f0d-4357-8d9a-5038a6bcb023.mp4
### Important Notes
**What is not implemented:**
* [Focus management between panels.](https://www.pivotaltracker.com/story/show/180872763) The grid is always focused. To accept the current input, use ctrl+Enter shortcut.
* [Proper handling of selection when having empty space on the right and pressing right arrow.](https://www.pivotaltracker.com/story/show/183487880)
* When entering a module, its name is not added to the input as described in the design doc. Will be a part of [this User Story](https://www.pivotaltracker.com/story/show/181058321).
**Known issues**
* [the selection, especially in the local scope section, has sometimes an undesirable offset](https://www.pivotaltracker.com/story/show/183487730). The cause is known, but not so easy to fix.
* The inserted nodes are often producing errors. The Browser's inherits the outdated understanding of the language from old Node Searcher, and it does not include new form of imports, static methods etc. Those all will be fixed as a part of [this User Story](https://www.pivotaltracker.com/story/show/181058321).
* The performance is improved, but still not ideal, due to problems in [text areas](https://www.pivotaltracker.com/story/show/183406745).
* To scroll the documentation panel, you must first click on it.
This change brings by-type pattern matching to Enso.
One can pattern match on Enso types as well as on polyglot types.
For example,
```
case x of
_ : Integer -> ...
_ : Text -> ...
_ -> ...
```
as well as Java's types
```
case y of
_ : ArrayList -> ...
_ : List -> ...
_ : AbstractList -> ...
_ -> ..
```
It is no longer possible to match a value with a corresponding type constructor.
For example
```
case Date.now of
Date -> ...
```
will no longer match and one should match on the type (`_ : Date`) instead.
```
case Date of
Date -> ...
```
is fine though, as requested in the ticket.
The change required further changes to `type_of` logic which wasn't dealing well with polyglot values.
Implements https://www.pivotaltracker.com/story/show/183188846
# Important Notes
~I discovered late in the game that nested patterns involving type patterns, such as `Const (f : Foo) tail -> ...` are not possible due to the old parser logic.
I would prefer to add it in a separate PR because this one is already getting quite large.~ This is now supported!
When nodes get invalidated in the cache, they have to be recomputed. Let the IDE know which of the nodes are pending by sending `Api.ExpressionUpdate.Payload.Pending` message.
# Important Notes
This PR introduces new `Api.ExpressionUpdate.Payload.Pending` message. This message is delivered before re-computation of nodes. Later `Api.ExpressionUpdate.Payload.Value` or other is sent to notify the IDE that a value for given node is available.
Trivial implementation of of the `Api.ExpressionUpdate.Payload.Pending` message in the IDE is provided by this PR to (improperly) visualize pending node status - further improvements needed in follow up PRs.
Makes statics static. A type and its instances have different methods defined on them, as it should be. Constructors are now scoped in types, and can be imported/exported.
# Important Notes
The method of fixing stdlib chosen here is to just not. All the conses are exported to make all old code work. All such instances are marked with `TODO Dubious constructor export` so that it can be found and fixed.
This change implements a simple `type_of` method that returns a type of a given value, including for polyglot objects.
The change also allows for pattern matching on various time-related instances. It is a nice-to-have on its own, but it was primarily needed here to write some tests. For equality checks on types we currently can't use `==` due to a known _feature_ which essentially does wrong dispatching. This will be improved in the upcoming statics PR so we agreed that there is no point in duplicating that work and we can replace it later.
Also, note that this PR changes `Meta.is_same_object`. Comparing types revealed that it was wrong when comparing polyglot wrappers over the same value.
This PR adds a possibility to generate native-image for engine-runner.
Note that due to on-demand loading of stdlib, programs that make use of it are currently not yet supported
(that will be resolved at a later point).
The purpose of this PR is only to make sure that we can generate a bare minimum runner because due to lack TruffleBoundaries or misconfiguration in reflection config, this can get broken very easily.
To generate a native image simply execute:
```
sbt> engine-runner-native/buildNativeImage
... (wait a few minutes)
```
The executable is called `runner` and can be tested via a simple test that is in the resources. To illustrate the benefits
see the timings difference between the non-native and native one:
```
>time built-distribution/enso-engine-0.0.0-dev-linux-amd64/enso-0.0.0-dev/bin/enso --no-ir-caches --in-project test/Tests/ --run engine/runner-native/src/test/resources/Factorial.enso 6
720
real 0m4.503s
user 0m9.248s
sys 0m1.494s
> time ./runner --run engine/runner-native/src/test/resources/Factorial.enso 6
720
real 0m0.176s
user 0m0.042s
sys 0m0.038s
```
# Important Notes
Notice that due to a [bug in GraalVM](https://github.com/oracle/graal/issues/4200), which is already fixed in 22.x, and us still being on 21.x for the time being, I had to add a workaround to our sbt build to build a different fat jar for native image. To workaround it I had to exclude sqlite jar. Hence native image task is on `engine-runner-native` and not on `engine-runner`.
Will need to add the above command to CI.
Implements https://www.pivotaltracker.com/story/show/183082087
# Important Notes
- Removed unnecessary invocations of `Error.throw` improving performance of `Vector.distinct`. The time of the `add_work_days and work_days_until should be consistent with each other` test suite came down from 15s to 3s after the changes.
Turns that if you import a two-part import we had special code that would a) add Main submodule b) add an explicit rename.
b) is problematic because sometimes we only want to import specific names.
E.g.,
```
from Bar.Foo import Bar, Baz
```
would be translated to
```
from Bar.Foo.Main as Foo import Bar, Baz
```
and it should only be translated to
```
from Bar.Foo.Main import Bar, Baz
```
This change detects this scenario and does not add renames in that case.
Fixes [183276486](https://www.pivotaltracker.com/story/show/183276486).
IR cache never really took into account a situation when a binding from the imported module has changed. In other words, it would continue to happily use the serialized metadata without noticing that it changed.
This change forces cache invalidation when any of the imported modules was invalidated (or rather not loaded from cache).
# Important Notes
Added simple test infrastructure that simulates file modifications that would trigger the initial cache invalidation.
If they succeed, cache invalidation is propagated thus causing an error.
`Vector` type is now a builtin type. This requires a bunch of additional builtin methods for its creation:
- Use `Vector.from_array` to convert any array-like structure into a `Vector` [by copy](f628b28f5f)
- Use (already existing) `Vector.from_polyglot_array` to convert any array-like structure into a `Vector` **without** copying
- Use (already existing) `Vector.fill 1 item` to create a singleton `Vector`
Additional, for pattern matching purposes, we had to implement a `VectorBranchNode`. Use following to match on `x` being an instance of `Vector` type:
```
import Standard.Base.Data.Vector
size = case x of
Vector.Vector -> x.length
_ -> 0
```
Finally, `VectorLiterals` pass that transforms `[1,2,3]` to (roughly)
```
a1 = 1
a2 = 2
a3 = 3
Vector (Array (a1,a2, a3))
```
had to be modified to generate
```
a1 = 1
a2 = 2
a3 = 3
Vector.from_array (Array (a1, a2, a3))
```
instead to accomodate to the API changes. As of 025acaa676 all the known CI checks passes. Let's start the review.
# Important Notes
Matching in `case` statement is currently done via `Vector_Data`. Use:
```
case x of
Vector.Vector_Data -> True
```
until a better alternative is found.
The goal of this request is to simplify hello world and other trivial Enso programs. No need to learn any standard library functions, enough to write:
```
main = "Hello World!"
```
and the result is going to be printed:
```bash
enso$ ./built-distribution/enso-engine-0.0.0-dev-linux-amd64/enso-0.0.0-dev/bin/enso --run hello.enso
'Hello World!'
```
the result is only printed, if it is not `Nothing`. E.g. if the last statement is `IO.print ...` (which returns `Nothing`), then no value is printed at the end by the launcher.
- Added `databases`, `database`, `set_database`.
- Added `schemas`, `schema`, `set_schema`.
- Added `table_types`,
- Added `tables`.
- Moved the vast majority of the connection work into a lower level `JDBC_Connection` object.
- `Connection` represents the standard API for database connections and provides a base JDBC implementation.
- `SQLite_Connection` has the `Connection` API but with custom `databases` and `schemas` methods for SQLite.
- `Postgres_Connection` has the `Connection` API but with custom `set_database`, `databases`, `set_schema` and `schemas` methods for Postgres.
- Updated `Redshift` - no public API change.
Implements https://www.pivotaltracker.com/story/show/182307143
# Important Notes
- Modified standard library Java helpers dependencies so that `std-table` module depends on `std-base`, as a provided dependency. This is allowed, because `std-table` is used by the `Standard.Table` Enso module which depends on `Standard.Base` which ensures that the `std-base` is loaded onto the classpath, thus whenever `std-table` is loaded by `Standard.Table`, so is `std-base`. Thus we can rely on classes from `std-base` and its dependencies being _provided_ on the classpath. Thanks to that we can use utilities like `Text_Utils` also in `std-table`, avoiding code duplication. Additional advantage of that is that we don't need to specify ICU4J as a separate dependency for `std-table`, since it is 'taken' from `std-base` already - so we avoid including it in our build packages twice.
This is a step towards the new language spec. The `type` keyword now means something. So we now have
```
type Maybe a
Some (from_some : a)
None
```
as a thing one may write. Also `Some` and `None` are not standalone types now – only `Maybe` is.
This halfway to static methods – we still allow for things like `Number + Number` for backwards compatibility. It will disappear in the next PR.
The concept of a type is now used for method dispatch – with great impact on interpreter code density.
Some APIs in the STDLIB may require re-thinking. I take this is going to be up to the libraries team – some choices are not as good with a semantically different language. I've strived to update stdlib with minimal changes – to make sure it still works as it did.
It is worth mentioning the conflicting constructor name convention I've used: if `Foo` only has one constructor, previously named `Foo`, we now have:
```
type Foo
Foo_Data f1 f2 f3
```
This is now necessary, because we still don't have proper statics. When they arrive, this can be changed (quite easily, with SED) to use them, and figure out the actual convention then.
I have also reworked large parts of the builtins system, because it did not work at all with the new concepts.
It also exposes the type variants in SuggestionBuilder, that was the original tiny PR this was based on.
PS I'm so sorry for the size of this. No idea how this could have been smaller. It's a breaking language change after all.
Avoid long lines when using `dom().set_inner_text` - rather split the long lines to 1024 chunks and insert them as individual `<div>` elements.
# Important Notes
I was testing the behavior on following program:
```
from Standard.Base import all
import Standard.Base.Data.Statistics
import Standard.Visualization
main =
number1 = 200000
operator1 = 0.up_to number1 . to_vector . map .noise
operator2 = operator1.sort
operator3 = operator2.to_text
```
before my change the visualization of `operator3` was blank. With my change it gets filled with data.
- Added `Zone`, `Date_Time` and `Time_Of_Day` to `Standard.Base`.
- Renamed `Zone` to `Time_Zone`.
- Added `century`.
- Added `is_leap_year`.
- Added `length_of_year`.
- Added `length_of_month`.
- Added `quarter`.
- Added `day_of_year`.
- Added `Day_Of_Week` type and `day_of_week` function.
- Updated `week_of_year` to support ISO.
# Important Notes
- Had to pass locale to formatter for date/time tests to work on my PC.
- Changed default of `week_of_year` to use ISO.
Implements https://www.pivotaltracker.com/story/show/182879865
# Important Notes
Note that removing `set_at` still does not make our arrays fully immutable - `Array.copy` can still be used to mutate them.
* Builtin Date_Time, Time_Of_Day, Zone
Improved polyglot support for Date_Time (formerly Time), Time_Of_Day and
Zone. This follows the pattern introduced for Enso Date.
Minor caveat - in tests for Date, had to bend a lot for JS Date to pass.
This is because JS Date is not really only a Date, but also a Time and
Timezone, previously we just didn't consider the latter.
Also, JS Date does not deal well with setting timezones so the trick I
used is to first call foreign function returning a polyglot JS Date,
which is converted to ZonedDateTime and only then set the correct
timezone. That way none of the existing tests had to be changes or
special cased.
Additionally, JS deals with milliseconds rather than nanoseconds so
there is loss in precision, as noted in Time_Spec.
* Add tests for Java's LocalTime
* changelog
* Make date formatters in table happy
* PR review, add more tests for zone
* More tests and fixed a bug in column reader
Column reader didn't take into account timezone but that was a mistake
since then it wouldn't map to Enso's Date_Time.
Added tests that check it now.
* remove redundant conversion
* Update distribution/lib/Standard/Base/0.0.0-dev/src/Data/Time.enso
Co-authored-by: Radosław Waśko <radoslaw.wasko@enso.org>
* First round of addressing PR review
* don't leak java exceptions in Zone
* Move Date_Time to top-level module
* PR review
Co-authored-by: Radosław Waśko <radoslaw.wasko@enso.org>
Co-authored-by: Jaroslav Tulach <jaroslav.tulach@enso.org>
Use Proxy_Polyglot_Array as a proxy for polyglot arrays, thus unifying
the way the underlying array is accessed in Vector.
Used the opportunity to cleanup builtin lookup, which now actually
respects what is defined in the body of @Builtin_Method annotation.
Also discovered that polyglot null values (in JS, Python and R) were leaking to Enso.
Fixed that by doing explicit translation to `Nothing`.
https://www.pivotaltracker.com/story/show/181123986
First of all this PR demonstrates how to implement _lazy visualization_:
- one needs to write/enhance Enso visualization libraries - this PR adds two optional parameters (`bounds` and `limit`) to `process_to_json_text` function.
- the `process_to_json_text` can be tested by standard Enso test harness which this PR also does
- then one has to modify JavaScript on the IDE side to construct `setPreprocessor` expression using the optional parameters
The idea of _scatter plot lazy visualization_ is to limit the amount of points the IDE requests. Initially the limit is set to `limit=1024`. The `Scatter_Plot.enso` then processes the data and selects/generates the `limit` subset. Right now it includes `min`, `max` in both `x`, `y` axis plus randomly chosen points up to the `limit`.
![Zooming In](https://user-images.githubusercontent.com/26887752/185336126-f4fbd914-7fd8-4f0b-8377-178095401f46.png)
The D3 visualization widget is capable of _zooming in_. When that happens the JavaScript widget composes new expression with `bounds` set to the newly visible area. By calling `setPreprocessor` the engine recomputes the visualization data, filters out any data outside of the `bounds` and selects another `limit` points from the new data. The IDE visualization then updates itself to display these more detailed data. Users can zoom-in to see the smallest detail where the number of points gets bellow `limit` or they can select _Fit all_ to see all the data without any `bounds`.
# Important Notes
Randomly selecting `limit` samples from the dataset may be misleading. Probably implementing _k-means clustering_ (where `k=limit`) would generate more representative approximation.
So far, when opening the searcher with no node selected, an empty input without an associated AST node was created. This input was manipulated by the user and the final expression from the input was used to create a new node when the user confirmed their input. This PR changes this, so that when the searcher is opened without a selected node, a new AST node is created right away with some placeholder content, and this node is updated when the user confirms their input.
The only change visible to the user, is that if the text editor is opened during editing, a new node will appear in the source when the searcher is opened to create a new node. All other behaviour should stay the same.
This change adds support for matching on constants by:
1) extending parser to allow literals in patterns
2) generate branch node for literals
Related to https://www.pivotaltracker.com/story/show/182743559
This change adds Autosave action for open buffers. The action is scheduled
after every edit request and is cancelled by every explicit save file request, if
necessary. Successful autosave also notifies any active clients of the buffer.
Related to https://www.pivotaltracker.com/story/show/182721656
# Important Notes
WIP
PR allows to attach metod pointers as a visualization expressions. This way it allows to attach a runtime instrument that enables caching of intermediate expressions.
# Important Notes
ℹ️ API is backward compatible.
To attach the visualization with caching support, the same `executionContext/attachVisualisation` method is used, but `VisualisationConfig` message should contain the message pointer.
While `VisualisationConfiguration` message has changed, the language server accepts both new and old formats to keep visualisations working in IDE.
#### Old format
```json
{
"executionContextId": "UUID",
"visualisationModule": "local.Unnamed.Main",
"expression": "x -> x.to_text"
}
```
#### New format
```json
{
"executionContextId": "UUID",
"expression": {
"module": "local.Unnamed.Main",
"definedOnType": "local.Unnamed.Main",
"name": "encode"
}
}
```
Importing individual methods didn't work as advertised because parser
would allow them but later drop that information. This slipped by because we never had mixed atoms and methods in stdlib.
# Important Notes
Added some basic tests but we need to ensure that the new parser allows for this.
@jdunkerley will be adding some changes to stdlib that will be testing this functionality as well.