Debugging the issue reported by @PabloBuchu when the language server initialization hangs in the cloud. I'm still not sure what is happening in the cloud because I was not able to reproduce it when trying to connect two clients simultaneously.
Another potential source of the issue may be the Scala Future -> Java CompletableFuture conversion, but I didn't find anything suspicious there.
The change upgrades `directory-watcher` library, hoping that it will fix the problem reported in #7695 (there has been a number of bug fixes in MacOS listener since then).
Once upgraded, tests in `WatcherAdapterSpec` because the logic that attempted to ensure the proper initialization order in the test using semaphore was wrong. Now starting the watcher using `watchAsync` which only returns the future when the watcher successfully registers for paths. Ideally authors of the library would make the registration bit public
(3218d68a84/core/src/main/java/io/methvin/watcher/DirectoryWatcher.java (L229C7-L229C20)) but it is the best we can do so far.
Had to adapt to the new API in PathWatcher as well, ensuring the right order of initialization.
Should fix#7695.
Added Blank_Selector constructor and applied to remove_blank_columns, select_blank_columns, filter_blank_rows for #7931 . Changed when_any to when for readability.
Previously, constant columns were given generated names with UUIDs in them, which are long and provide no information. Instead, we now use the constant value itself to form the name.
Since these new generated names are less unique, we must explicitly make them unique, in cases where the caller did not explicilty set a name.
While working on #8158, I noticed a crash when the `data` directory is created at the project root. Turns out it is the issue in the `ydoc-server`, which thinks every filesystem event is about files. Unfortunately, we don‘t have the needed info available, so we need to make the `file/info` request.
- Closes#7981
- Adds a `RUNTIME_ERROR` operation into the DB dialect, that may be used to 'crash' a query if a condition is met - used to validate if `lookup_and_replace` invariants are still satisfied when the query is materialized.
- Removes old `Table_Helpers.is_table` and `same_backend` checks, in favour of the new way of checking this that relies on `Table.from` conversions, and is much simpler to use and also more robust.
Fixes#8186 by turning `IllegalStateException` into log message. Re-assigning of `BindingsMap` can happen in the IDE where evaluation of modules is repeated again and again. In addition to that avoid dropping errors in compiler without them being noticed.
Using a `TruffleLogger` in `SerializationManager` that is bound to the engine rather than the context prevents reaching an illegal state when using thread pools.
Also cleaned up some tests for consistency.
To verify the fix
```diff
--- a/engine/runtime/src/main/scala/org/enso/compiler/SerializationManager.scala
+++ b/engine/runtime/src/main/scala/org/enso/compiler/SerializationManager.scala
@@ -31,7 +31,7 @@ final class SerializationManager(compiler: Compiler) {
import SerializationManager._
/** The debug logging level. */
- private val debugLogLevel = Level.FINE
+ private val debugLogLevel = Level.INFO
```
and run
`sbt:enso> runtime/test`
Closes#8147.
The qualified name for nodes was incorrect, resulting in the suggestion entry never being found.
# Important Notes
Should double-check that the issue is present on develop but not this PR. I've checked locally but it doesn't hurt to be extra sure.
Implementing most parts of the Documentation Panel in the new GUI.
Known issues:
- Links do not work (yet, covered by #7992)
- Some pages are entirely empty – I’m investigating. Missing doc sections likely cause this, so we probably want to add some placeholder.
- Tags do not look as in design. I tried implementing them correctly but didn't have enough time to finish everything. Some initial implementation is in place, though. I will create a separate task for them.
https://github.com/enso-org/enso/assets/6566674/a656ae78-5d4c-45f4-a0a5-e07fa573253e
The change eliminates a race-condition that can appear between the `PathWatcher` deregistering the last client (and shutting down) and `ReceivesTreeUpdatesHandler` receiving the termination message of that event. In between there could come a message towards the mentioned `PathWatcher` resulting in a timeout.
The scenario has become rather common in CI tests resulting in spurious failures. The problem could also be simulated by adding artificial `Thread.sleep` between sending the reply in `PathWatcher` and stopping the actor.
Fixes#8151.
# Important Notes
Example failure https://github.com/enso-org/enso/actions/runs/6642894609/job/18048711561?pr=8145
To simulate the scenario
```diff
diff --git a/engine/language-server/src/main/scala/org/enso/languageserver/filemanager/PathWatcher.scala b/engine/language-server/src/main/scala/org/enso/languageserver/filemanager/PathWatcher.scala
index 713cfe1182..88afac95cb 100644
--- a/engine/language-server/src/main/scala/org/enso/languageserver/filemanager/PathWatcher.scala
+++ b/engine/language-server/src/main/scala/org/enso/languageserver/filemanager/PathWatcher.scala
@@ -110,6 +110,7 @@ final class PathWatcher(
case UnwatchPath(client) =>
sender() ! CapabilityReleased
+ Thread.sleep(2000)
unregisterClient(root, base, clients - client)
```
Towards reduced reliance on Scala semantics.
Translated IR.scala to IR.java and extracted implicits that now need to be imported explicitly.
# Important Notes
1:1 translation. For now `@Identifier` and `@ExternalID` represent the old type aliases but are not verified at compile time.
This is because in a mixed Scala/Java world this seems impossible to employ such frameworks as Checker.
[vokoscreenNG-2023-10-23_06-35-31.webm](https://github.com/enso-org/enso/assets/1047859/d8a5ffc5-4677-4264-a246-8def4418505a)
- Implement disconnected edges and edge interactions (#7864).
- Port edge layout from Rust.
- Implement output ports, hover extensions, backward-edge arrows, hover-splitting.
# Important Notes
- A new `currentInteraction` API supports mutually-exclusive interactions; this is used to cancel edge creation when the CB is opened. We can use it more generally to ensure only one major interaction is ongoing at a time.
- Remaining details to reach parity with gui1 edges: #8139
`capability/acquire` with an invalid method `executionContext/canModify` would previously timeout because the capability router simply didn't support that kind of capability request.
Now rather than timing out, indicating a failure on LS, we report an error.
Closes#8038.
After a discussion, I was really curious that our panics are supposed to be almost free - and while trusting that statement, it was really hard to believe - so I wanted to see for myself - knowing that an experiment is the most robust source of this kind of information - testing that in practice.
So I wrote a benchmark comparing various ways of reporting errors, also testing them both at 'shallow' and 'deep' stack traces (adding 200 additional frames) - to see how stack depth affects them, if at all.
The panics are indeed blazing fast! Kudos to the engine team. However, it seems that our dataflow errors are relatively slow (and we tend to use them _more_ than panics and want to be using them more and more). This uncovers a possible optimization opportunity. Can we make them as fast as panics??
Analysis of the benchmark results in comment below.
In order to execute `runtime-parser` in browser, we need to slightly simplify dependencies of our code.
# Important Notes
The actual PR explaining why these changes are desirable is #8094
Disables visual tests, because they are occasionally flaky (e.g. the recent issues with lint CI action).
This is very bad as it would cause every PR to (potentially) cause other PRs to fail, if the flaky test is not caught in the CI runs in the offending PR.
# Important Notes
None
- Fixes the issue that sometimes occurred on CI where old `services` configuration was not cleaned and SPI definitions were leaking between PRs, causing random failures:
```
❌ should allow selecting table rows based on a boolean column
An unexpected panic was thrown: java.util.ServiceConfigurationError: org.enso.base.file_format.FileFormatSPI: Provider org.enso.database.EnsoConnectionSPI not found
```
- The issue is fixed by detecting unknown SPI classes before the build, and if such classes are detected, cleaning the config and forcing a rebuild of the given library to ensure consistency of the service config.
- Closes https://github.com/enso-org/cloud-v2/issues/721
- Drag-n-drop for adding labels to an asset
- Adds to a single asset when hovering over an asset that is not selected
- Adds to all selected assets when hovering over an asset that is selected
### Important Notes
- The backend endpoint for associating labels with assets is currently broken - specifically, it errors when the new list of assets is empty.
- Fixes#8049
- Adds tests for handling of Date_Time upload/download in Postgres.
- Adds tests for edge cases of handling of Decimal and Binary types in Postgres.
close#8033
Changelog:
- update: run language server initialization once
- fix: issues with async `getSuggestionDatabase` message handling in new IDE
- update: implement unique background jobs
- refactor: initialization logic to Java
- refactor: `UniqueJob` to a marker interface
Improved warning reporting by not reporting IR in user-directed warning
messages. Made sure that fresh names retain some knowledge of the name
which they were created from.
Closes#7963.
- Closes#8062
# Important Notes
- This PR uses a completely separate implementation from the regexes used for the existing filtering. AFAICT, this is unavoidable, since this needs to retrieve matched text.
- I've come across a couple minor bugs related to the Component Browser:
- When deleting text (backspacing), the full list of unfiltered entries momentarily reappear because the filtering is reset.
- Things like `h.m` match `HTTP_Method` - if this is not intentional, it seems like it might be more difficult to fix, unfortunately.
- Note that highlighting (I assume correctly?) *does not* highlight any part of `HTTP_Method`.
Unmasked position info for the failed edits so that it is possible to understand what edits are attempted at all. So far unable to locate the problem based solely on the logs and the stacktrace. Added a test confirming that making edits at the end of the file continues to be fine (that's where the problem usually manifests itself).
# Important Notes
References #7978
- Validate spans during existing lexer and parser unit tests, and in `enso_parser_debug`.
- Fix lost span info causing failures of updated tests.
# Important Notes
- [x] Output of `parse_all_enso_files.sh` is unchanged since before #7881 (modulo libs changes since then).
- When the parser encounters an input with the first line indented, it now creates a sub-block for lines at than indent level, and emits a syntax error (every indented block must have a parent).
- When the parser encounters a number with a base but no digits (e.g. `0x`), it now emits a `Number` with `None` in the digits field rather than a 0-length digits token.
