Fixes#7737
Added structures representing Suggestion Database entries. Currently, the db is loaded from a snapshot from the old GUI.
Added an input to CB and use it to filter components. The interpretation is simple: the input is split by the last dot, and the left part is considered a qualified name, and the right part is a function name written by the user so far. I rewrote the filtering algorithm designed by @jdunkerley, changing it a bit, so we support qualified names instead of just a type name.
https://github.com/enso-org/enso/assets/3919101/76a957f6-e53f-49ad-996c-398cd7112fc6
# Important Notes
* The component list is now sorted from "first to select" to "least interesting". The panel itself cares about putting the first on the bottom.
* The suggestion db snapshot is very big, so it's instead loaded from external server.
Closes#7677 by eliminating the _stackoverflow execption_. In general it seems _too adventurous_ to walk members of random foreign objects. There can be anything including cycles. Rather than trying to be too smart in these cases, let's just rely on `InteropLibrary.isIdentical` message.
# Important Notes
Calling `sort` on the `numpy` array no longer yields an error, but the array isn't sorted - that needs a fix on the Python side: https://github.com/oracle/graalpython/issues/354 - once it is in, the elements will be treated as numbers and the sorting happens automatically (without any changes in Enso code).
Resolve macros eagerly. Improves performance; allows parser to handle arbitrarily-long lines (fixes#7691).
# Important Notes
- A new utility, `lib/rust/parser/debug/tools/parse_all_enso_files.sh`, supports comparing ASTs parsed with different versions of the parser. This tool has been used to verify that this refactor doesn't change the result of parsing any standard library or test file.
- Fixes#7799
The PR that introduced this bug stopped event propagation in certain cases, in order to fix another issue.
This PR introduces a way to run an event handler, but indicate failure so further handlers (if any) are run, otherwise the event will be propagated all the way to the document root.
# Important Notes
None
close#7520
Changelog:
- update: SectionsToBinOp compiler pass produces function application for left sections
- refactor: simplify the registration of builtin methods
close#7765
Changelog:
- update: instead of relying on the connection closed events, the `sendSuggestionsDatabase` request initiates the suggestions database re-indexing
Add support for recovering from GL context loss. When the context is restored, the loading spinner is shown until shaders finish recompiling.
[vokoscreenNG-2023-08-25_09-39-11.webm](https://github.com/enso-org/enso/assets/1047859/cfa90ec5-72a1-41e6-bafa-177fa5e85fb2)
*While the context is missing, the loading spinner is rendered in the 0% state. (This condition will not normally be observed, except momentarily, as the browser should restore the context immediately if it is lost while the page is visible.) When we receive a new context, the spinner switches to the 90% state until restoration completes. Restoration is fast, as we don't need to do much work except recompiling shaders.*
# Important Notes
- A new debug hotkey, Ctrl+Alt+Shift+X, causes context loss for testing. Pressing it a second time causes context restoration.
- `Texture` is still a CPU-bound texture. It now uses the "immutable" `texStorage/texSubImage` API, which is a ["preferred alternative"](https://registry.khronos.org/webgl/specs/latest/2.0/#3.7.6) to the `texImage` API because it can be more efficient.
- The type for texture uniforms is now `Uniform<Option<Texture>>`. Texture uniforms are decoupled from the context.
- A new `ContextLost` error type can be returned by functions that cannot complete if the context is lost.
- Fix some crashes that could occur when context was lost.
- Clarify ownership of some rendering-related types: Externalize, and where possible eliminate, `Rc/RefCell`s.
Adds widgets:
- Checkbox (with sorting)
- Numeric slider
- Dropdown (accepting a list of strings)
- Closes#7731
- Placeholder (underscore - has no actions)
# Important Notes
The widgets are currently added to every node, but are not synced with the yjs representation. This is intentional, as (afaict) the format for the AST representation is not yet finalized.
There are a number of design differences, for practical reasons:
- The dropdown now has a scrollbar.
- As a side effect, the sort button needed to be moved left, to avoid overlapping with the scrollbar.
- Note that it is *not* centered in the 8px horizontal padding. It is 4px wide, and has 1px left and 3px right padding: `.||||...`. (Note that the 8px horizontal padding from the design is retained.
- 4px of vertical padding has been inserted, so that there is *some* padding between the bubble for the selected item, and the outer dropdown container, when the first item is selected. Note that this is different to the 8px
- 16px of right margin has been inserted after every item. This is the same amount of padding that is added by the bubble. This means that the dropdown does not change in width when a long item is selected.
Design issues:
- The sort button for the dropdown overlaps the text
close#7608
Changelog:
- update: log separately the evaluation of visualization expression and its arguments
- update: add visualization expression, its arguments, and the value type to the log
# Important Notes
Example
```
[TRACE] [2023-09-06T20:41:45+03:00] [enso] Executing visualization [VisualizationConfiguration(d195fdd8-d4e8-400f-a0d4-e50417eddd0a,ModuleMethod(MethodPointer(Standard.Visualization.Table.Visualization,Standard.Visualization.Table.Visualization,prepare_visualization),Vector(1000)),local.New_Project_1.Main)] on expression [ddc060df-9b59-48e5-bc61-aca849347343] of [class org.enso.interpreter.runtime.data.vector.Vector$Generic]...
```
- Fixes#7354
- And also closes#7712
- Refactors how we handle numeric ops - ensuring that the 'kernels' are placed all in one place and selected based on storage types.
Fixes the _RefactoringTests - rename project_ part of the #7775
Changelog:
- update: send the ok response before the notification to fix the order of events in tests
This PR addresses two problems mentioned in #7766:
1. A random integer overflow, likely caused by a bug in Rust parser
2. A concurrent access to a methods' map
Re 1: Unable to reproduce but it doesn't mean it won't happen again. Added a try/catch to get in the logs source code that caused it **and** not crash hard when it occurs.
Re 2: Changing methods map from `HashMap` to `ConcurrentHashMap`. Due to a poor design we leaked the underlying structure in a number of places, unnecessairly. `ConcurrentHashMap` does not accept `null` keys therefore due to leaking implementation had to ensure that `methods` of `ModuleScope` never escapes as-is.
Both workarounds should ensure that we don't crash hard when they appear.
Closes#7766
Updated the Component Browser stub from @Frizi's branch. Filled with some mocked data and implemented highlight and navigation.
https://github.com/enso-org/enso/assets/3919101/6af57413-bcb6-4402-9a4a-a992d58504ba
# Important Notes
Still missing the input panel, and obviously, the integration with anything will be a part of another task.
- Closes#7730. Adds top bar containing:
- Project title
- Execution mode selector (design/live)
- Navbar containing:
- Navigation previous/next button
- Breadcrumbs
# Important Notes
Icons are currently all separate files. This may need to be changed to a single svg containing individual icons in `defs` if(/when) that is merged into `develop`.
- Renames "Move To Trash" to "Delete" on local backend
- This is because the local backend does not have a trash category (or categories at all)
- Fix search bar autofocus
- The "delete" hotkey was focusing the search bar - this was because the shortcuts handler, and the autofocus handler, were both registered on the same `EventTarget`. The fix is to do `stopImmediatePropagation()` to stop running any other event handler, even ones on the same `EventTarget`.
# Important Notes
None
- Fixes https://github.com/enso-org/cloud-v2/issues/659
`getProjectDetails` was calling `listVersions` to get a fallback IDE version, creating a lot of unnecessary requests to the backend, *especially* because `getProjectDetails` is called periodically when a project is opening.
In this PR, the implementation has been changed to cache the fallback version for one day, meaning the extra `listVersions` calls should now only ever happen once per client per day.
# Important Notes
None
- Closes https://github.com/enso-org/cloud-v2/issues/619
- Async execution controls
# Important Notes
There is no design for this, as such, implementation details use placeholder designs.
- The context menu uses a play icon. An icon similar in style to the "copy" icon *may* work to represent "run in background", but it may be difficult to visually represent that it is being run in the background, without obstructing it with a foreground window
- The icon for projects being run in the background have a green tint, to distinguish them from projects that will be (or are currently) opened in the editor.
- this will ***almost certainly*** need to be replaced with a proper design
- This *may* also make sense for the local backend, *however* as I don't know whether there is a way to access the completion progress of execution from the PM API, local backend support is currently *not* implemented in this PR.
- On a related note: as far as I am aware, there is also no such endpoint for the cloud backend. However, support for async execution was recently added, so I am adding the basic functionality corresponding to the `executeAsync` project state.
- Whether a project is being run in the background is currently lost on refresh. This is because the async execution state is currently not sent by the backend.
- Placeholder shortcuts have been added (Shift+Enter - Cmd+Enter is already taken by the "share" action, and shift+double click). These are totally optional, and can easily be removed.
- Closes#7238
- Aligns `update_database_table` to a more consistent and clearer API - `update_rows`.
- Adds a `truncate_table` helper function, to pair up with `drop_table`. Both are `PRIVATE` for now.
- Adds tests for NULLs in keys in `update_rows` and `delete_rows`.
- The behaviour is sometimes unexpected, so instead these fail with `Null_Values_In_Key_Columns`.
- Adds a workaround for https://github.com/oracle/graal/issues/7359
- Adds a workaround for a related bug where a stack frame has no name (its `rootNode.getName() == null`).
- I could not track down this bug to provide a neat repro.
Closes: #7309
The changed origin of the visualization container caused the issue. Now, the origin is at the top left corner of the visualization, and bounding box abstraction expects the origin at the bottom left. Despite the comments in the code, the bounding box is designed to work with bottom-left origin only.
https://github.com/enso-org/enso/assets/6566674/273abfaa-45b6-4374-8d4c-3b8c4e2c1fc2
Again I hit the problem with shader compilation failing with no apparent reason, and again bumping electron version helped me.
# Important Notes
I bumped to latest `25.x` version as I haven't dared to change major version expecting breaking changes.
While looking into #7698 discovered that the `buildGraalDistribution`
task would fail to use the old (pre-23.0) GraalVM naming of artifacts.
This has changed since #7176.
The fix does not attempt to fix a problem of packaged GraalVM name in
`runtime`. That, as the ticket mentiones it, is the role of rust build
script that creates the correct bundle.
