[2023-12-12] Team GUI Demo Day

[farmaazon]

• The main focus were E2E Tests in CI #8462 (finished, but still facing some CI issues)
  • The build script ./run gui2 test - runs both unit and e2e tests in console
  • For debugging, one can run npm --workspace=enso-gui2 run test:e2e -- --ui with nice interface
  • There is also VSCode plugin, but I haven't used it yet.

Besides, fixed several discovered regressions:

• Worked around an error in production builds preventing opening projects Fix package #8458
• The not working visualizations Fix visualizations #8479
• Editing by click: Quickfix for editing problem #8456
• Some CB fixes #8507 - waiting for review :
  • Delete key not working in CB and in Code Editor
  • Cursor not placed at the end when adding new node with source node
  • Garbage added when edited node while having another one selected.
  • Closing CB when navigating.
  • Too eager selecting component after filtering update.
  • Premature node creation when dropping edges
  • Discarding changes when clicking-off CB

[MichaelMauderer]

Implemented entering/leaving nodes (#8492) + integration with breadcrumbs (#7788):

Peek.2023-12-07.13-28.webm

Peek.2023-12-12.10-46.webm

[vitvakatu]

• Fixing a couple of minor issues in dynamic dropdowns: Small fixes for GUI2 #8478
• A small fix for the old GUI: Sync presenter state not only on abort, but also on commit #8464
• Working on Collapsing nodes #8067 – waiting for AST edits.
• Fixing issues in the documentation panel: Improve documentation panel #8135

fixed-dropdown.mp4

documentation.improvements.mp4

[kazcw]

This past week:

• Ported new app features to the new AST, and merged the AST PR: New AST types #8263
• Added a new syntax in the parser (Parse inline function signatures #8470) and ensured support for an upcoming feature (default) works consistently between frontend/backend/parser.
• Implemented simpler AST match/replace based on new AST edit APIs, to support Setting evaluation context on a single node. #8072. [PR 1/2 for Ast-driven edits #8367; will be ready this morning.]

Next:

• Part 2/2 PR for Ast-driven edits #8367. Will eliminate the need to maintain an IdMap, and will make reordering blocks simple (see Reorder nodes after connecting. #8494).

[somebody1234]

Mostly various small fixes/other small QoL improvements this week.

Merged:

• Minor QoL improvement: Add new magnet axes #8506
• Minor fix: Fix node output type #8488
• Temporary workaround: Fix missing edges #8461
  • Long term plans according to the issue's comments is that PropertyAccess shouldn't contain nested Ident nodes, only nested Ident tokens
• Implemented old feature request: Read file directly from location #8490
• Fix on Dashboard side: Hide GUI2 when not in Editor tab #8445

Worked on:

• Minor restructure of @/util Cleanup @/util folder #8501
• Porting to new AST: Set output evaluation context for a single node #8440
• Fixes for downloading Enso Font in CI: Fixes for Enso Font in GUI2 #8508
• (Heavily WIP) Fixing crashes when calling openFile Fixes for Language Server sync server #8514

[mwu-tow]

I've done an extensive investigation of the NSIS installer issue.

Essentially, the issues are caused by the paths exceeding MAX_PATH, NSIS has trouble properly dealing with them. Also, diagnostics is really confusing, as the error handling code (and messages) are unclear or plain wrong.

The issue happens only on newer electron-builder versions, because a new update procedure was provided that first tries to backup the original installation by moving it away to a temporary location. This temporary location has a longer path (C:\Users\mwurb\AppData\Local\Temp\nsz758E.tmp\old-install vs C:\Users\mwurb\AppData\Local\Programs\Enso) which causes some paths to go over the limit.

However, we are already balancing on the limit. If we have a slightly longer path, the issue will also appear during initial installation. Whether we can reasonably avoid the problem on our side is being discussed in https://github.com/orgs/enso-org/discussions/8509

---

The alternative tools offered by electron-builder (and electron-forge) are not really suitable for us, as they either have similar issues or are not really maintained.

---

As an alternative approach, I'm investigating how feasible it would be to provide our own installation script. We already skip the vast majority of NSIS features (as we want one-click installation) and do not care about its other characteristics (like being only 35kB or supporting older Windows).

[Frizi]

• Finishing work on setting values for placeholder arguments (and unifying widget value updates in general). This is somewhat broken due to not having AST updates implemented yet. Accepting value on argument placeholders #8258
• Addressed incorrect whitespace rendering between widgets. Now spaces around operators will show up properly.

placeholder-updates.mp4

[JaroslavTulach]

2-3x test speedup by reduced memory usage

• Cleanup after each test run #8418

Initially

Now

Inline Execution Work

...continues...

Performance of the EnsoHashMap shall no longer be a problem with #8425

org_enso_benchmarks_generated_Enso_Hash_Map_100000_Enso_Incremental achieves 4.06ms per operation which makes it slightly faster than org_enso_benchmarks_generated_Enso_Hash_Map_100000_Java_Incremental benchmark.

org_enso_benchmarks_generated_Enso_Hash_Map_100000_Enso_Replacement has been sped up 400 times and is now just 50% slower than the org_enso_benchmarks_generated_Enso_Hash_Map_100000_Java_Incremental - that's a good result given that the Java HashMap doesn't need to keep the referential transparency at all.