Tabbing across widgets not working for newly added components about enso HOT 7 CLOSED

From my observation, this happens because once the year is entered, a placeholder widget is replaced with a non-placeholder version, so in effect it appears to stop being the currently active element.

In fact, the whole node view is replaced with a brand new one, because its top-level Ast element changes.

from enso.

Paweł Grabarz reports a new STANDUP for the provided date (2024-07-08):

Progress: Trying to separate Node IDs from AST IDs by assigning them sequentially and providing a stable cached mapping that can survive changes of either AST ID or External ID individually. It should be finished by 2024-07-15.

from enso.

Paweł Grabarz reports a new STANDUP for the provided date (2024-07-09):

Progress: Given up on stable mapping idea due to significant added complexity and potential desync issues if we ever need to broadcast per-client state (e.g. node selection of other clients). Instead decided to use Root expression External IDs for for Node IDs and ensure those are always preserved across all node edits. It should be finished by 2024-07-15.

from enso.

Paweł Grabarz reports a new STANDUP for yesterday (2024-07-10):

Progress: Refactored Node IDs to be based on External IDs, fixed External ID and metadata preservation in case of edits that deleted first filled-in argument of function application that was a root of the node. It should be finished by 2024-07-15.

from enso.

Paweł Grabarz reports a new STANDUP for today (2024-07-11):

Progress: Added many assertions around usages of NodeId/AstId/PortId, added more tests. Fixed and improved stability of existing e2e tests. Submitted PR for merging. It should be finished by 2024-07-15.

from enso.

Paweł Grabarz reports a new STANDUP for the provided date (2024-07-04):

Progress: Started investigating the issue. Found out that the widgets or whole nodes are being incorrectly rendered from scratch on certain node changes. It should be finished by 2024-07-15.

from enso.

Paweł Grabarz reports a new STANDUP for the provided date (2024-07-05):

Progress: Understood the root cause of issue, which is that Node IDs are not correctly preserved across some edits due to AST type changes at the root of their expression. Attempted to switch View layer to use different key calculation, but that led to many subtle bugs with incorrectly indexed node state. This will require a more in-depth solution. It should be finished by 2024-07-15.

from enso.