Epic: Editor & Modeling 3001

[EDsCODE]

Goal: We're undergoing a rebuild of our data warehouse/sql editor scene to improve UX and become a full fledged and joy-sparking editor.

MVP - this will be sufficient as a replacement for the existing editor

• new layout that full screens the editor with resizable panels (feat(data-warehouse): new SQL layout #25686)
• multi tab support (feat(data-warehouse): new SQL layout #25686)
• virtualized result table component (feat(data-warehouse): new SQL layout #25686)
• integrate nav 3000 sidebar with sources, tables, views (feat(editor-3000): integrate nav3000 sidebar #26184)
• sidebar is searchable (feat(editor-3000): integrate nav3000 sidebar #26184)
• refactor logic for multitab editor (chore(data-warehouse): refactor logic #26128)
• click on a schema and see the fields in sidebar (feat(editor-3000): integrate nav3000 sidebar #26184)
• sidebar sources section should link to source configuration (feat(editor-3000): integrate nav3000 sidebar #26184)
• Add visualization tools to result pane as a new tab (feat(editor-3000): add visualization tab #26268)
• Edit view feature (feat(editor-3000): integrate nav3000 sidebar #26184)
• Improvement: Export results view as CSV
• Improvement: Save views without aliasing (chore(data-warehouse): Saving as view shouldn't require aliasing every field #26219)
• query time metadata info line
• handle mobile nav bar
• reintegrate variables in editor (chore(editor-3000): reintegrate editor variables in editor 3000 #26587)
• autocomplete (chore(editor-3000): init all models for autocomplete #26626)
• notice banners are buggy as they're rendered behind the sidebar and above the editor
• don't load again selecting between tabs
• adding a bunch of tabs should scroll/not extend the page
• Editable tab naming (feat(rename-sql-editor-tabs): Allows renaming of tabs #27608)

Upcoming:

• Canvas view of table relationships (another tab alongside results/visualizations)
• don't have implicit LIMIT on queries in SQL editor
• Reorganize sidebar to enable drafts/views/tables @EDsCODE
• after ^ Draft query edits/version history (handling multiple people making changes)
• virtualize table component (on visualization option) or paginate the visualization table so it doesn't overwhelm
• scheduling data modeling runs
• Search should search tables and fields
• doc links
• Guidance around special syntax (links in tables)
• Joins: list of fields is disorganized and unintuitive
• Remove the editor from insights page (consider)
• adding descriptions/annotations to tables and fields context/idea

Further out:

• SQL autocomplete

[annaszell]

Some of my wishlist items:

• debugging --> I am prone to make silly mistakes (forgetting a comma, a parenthesis, etc.), so if the error message could tell me or maybe even highlight the row where the issue is, that would be simply lovely. I know we have the red underlining but it doesn't always work properly. iirc Databricks highlights the part of your code in red that results in it not running (I haven't used Databricks in over a year now though so not 100% sure)
• formatting --> I'm terrible at being consistent when I write my queries, so before I share it for review with someone, I usually copy-paste it to ChatGPT to format it for me, so people have an easier time understanding what's going on
• clearer indication that by default only the first 100 rows will be shown (e.g. when I open a new tab, the example query has the LIMIT 100 but if I remove that and rerun it, it's still 100 results), and how can I override this
• partial runs --> so especially when I'm investigating something, it would be helpful to just run parts of the query within the same tab by using ; at the end of each query instead of commenting things out. sometimes I forget the exact fields/properties and I just want to have a quick look at the schema, etc. e.g. below 👇 and I could just highlight the one I want to run, it works like this in Metabase (and others)

select
id,
country,
event
from events;

select * from events limit 1;
this is not a super annoying thing but it would be helpful in certain scenarios

• time outs --> this one is a minor issue but I'd love it if a query that would eventually time out wouldn't even run, just tell me it's not gonna happen so I don't sit there for minutes. I know it's a first world problem, hence being the last one