Commit Graph
9 Commits
Author SHA1 Message Date
6186a0645d feat: edit variables, resources and triggers in their own session tab (#11206)
* feat: edit variables, resources and triggers in their own session tab

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* fix: take a trigger's flow kind from its config and name the written trigger

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* refactor: one operating-workspace context for editors acting on a session's workspace

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* fix: reload page item tabs only in sessions acting on the tool's workspace

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* fix: resolve open in workspace for every preview tab kind in one place

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* fix: guard every component under a session editor against the navigation workspace

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* fix: show a page item tab whose item is gone instead of mounting its editor

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* fix: follow a saved page item in the tab's own workspace and restore in-frame rows

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* feat: close a page item tab from its editor header

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* fix: judge trigger permissions by the user acting in the session workspace

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* fix: judge every session-editor permission by the user acting in that workspace

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* fix: keep session permissions reactive to the acting role and restore tabs as last seen

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* fix: default Path's owner controls to the user acting in its workspace

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

---------

Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
Co-authored-by: Ruben Fiszel <ruben@windmill.dev>
2026-09-19 23:29:14 +02:00
386e115daa feat: lazily expand s3 explorer folders one level at a time (#10420)
* feat: wire paged object storage listing module

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JqAc8mz6Gu698kBbJJVwcT

* feat: document list_stored_files_paged endpoint in openapi

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JqAc8mz6Gu698kBbJJVwcT

* feat: lazily expand s3 explorer folders one level at a time

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JqAc8mz6Gu698kBbJJVwcT

* chore: pin ee-repo-ref to the paged listing branch

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JqAc8mz6Gu698kBbJJVwcT

* fix: share object_store credential resolution and surface listing errors

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Sqq2LhmWaGwP11Cf3UqWxe

* Chevron is cool

* page size 5000

* feat: make the load more row full-width, secondary and chevron-led

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Sqq2LhmWaGwP11Cf3UqWxe

* fix: render newly loaded flat pages inside already-expanded folders

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JqAc8mz6Gu698kBbJJVwcT

* fix: address review findings in the lazy s3 explorer

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JqAc8mz6Gu698kBbJJVwcT

* fix: address review nits in the lazy s3 explorer

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JqAc8mz6Gu698kBbJJVwcT

* chore: bump ee-repo-ref after merging main

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JqAc8mz6Gu698kBbJJVwcT

* fix: document ambient credential contract and constrain max_keys schema

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JqAc8mz6Gu698kBbJJVwcT

* fix: treat an exhausted page token as exhausted, not as a continuation

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JqAc8mz6Gu698kBbJJVwcT

* chore: bump ee-repo-ref for canonical prefix validation

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JqAc8mz6Gu698kBbJJVwcT

* chore: bump ee-repo-ref for prefix scoping and opaque cursors

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JqAc8mz6Gu698kBbJJVwcT

* fix: invalidate a folder's in-flight load when deleting from it

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JqAc8mz6Gu698kBbJJVwcT

* fix: discard a stale folder page after its level is invalidated

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JqAc8mz6Gu698kBbJJVwcT

* chore: bump ee-repo-ref for bounded local listing

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JqAc8mz6Gu698kBbJJVwcT

* fix: label folders whose final path segment is empty

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JqAc8mz6Gu698kBbJJVwcT

* feat: search files by any part of their path, not just folder prefix

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JqAc8mz6Gu698kBbJJVwcT

* feat: search files by path prefix instead of a full-bucket scan

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JqAc8mz6Gu698kBbJJVwcT

* fix: guard stale search responses and describe prefix search accurately

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JqAc8mz6Gu698kBbJJVwcT

* chore: bump ee-repo-ref for the search prefix fallback fix

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JqAc8mz6Gu698kBbJJVwcT

* chore: regenerate the served openapi specs

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JqAc8mz6Gu698kBbJJVwcT

* chore: bump ee-repo-ref for the search cursor fallback fix

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JqAc8mz6Gu698kBbJJVwcT

* chore: bump ee-repo-ref for the bounded search scan

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* fix: surface a failed flat listing instead of spinning forever

The flat branch of loadFiles was awaited without a catch, and loadFlatFiles
clears its loading flags only on the success tail. Every caller reaches it
un-awaited, so a rejected listing left the drawer on "Loading content" with
nothing reported. Routing the filter box through this arm made it reachable
per keystroke rather than once per open.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* fix: give back the flat cursor when a page fails to load

"Load more" advanced `page` before requesting it, so a failed page left the
cursor pointing at a `listMarkers` slot that was never filled. The retry sent
no marker at all and silently replayed the first page, and the
`listMarkers.length == page` guard kept it there until the listing was reset.

Only reachable now that a failed page is retryable rather than a permanent
spinner.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* fix: scope the flat cursor rollback to its own listing

The rollback matched on the page number alone, so a page that failed after a
filter or storage change could roll back the *replacement* listing once it had
reached the same number, stranding its cursor. Tie it to the generation the
request was issued under.

The delete replay loop had the mirrored problem: it re-drove `page` by hand and
carried on past a failed page, leaving `page` ahead of `listMarkers` for good.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* fix: skip the delete replay when the fresh listing itself failed

clearAndLoadFiles dropped the result it already computes, so a failed
post-delete listing still ran the replay loop: each page advanced `page` with
an empty `listMarkers`, which never recovers because the marker-length guard
only pushes when the two agree. Every later "Load more" then replayed page one.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* fix: stop a superseded lazy load from writing into the search that replaced it

loadFolderPage resolves rather than throwing once its generation is stale, so a
filter change that switches the picker to the flat listing mid-flight left the
lazy branch free to expand a preselected file into the search's results and to
clear the search's loading flags. Guard both on the generation it started under,
as the flat branch already does.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* fix: check the listing generation throughout the reveal walk

Revealing a preselected key is a chain of round trips, so checking once at entry
left the rest of the walk free to keep loading after a filter change had already
switched the picker to the search — under the replacement generation, so the
per-level guards inside loadFolderPage saw nothing wrong.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* fix: let a late metadata failure clear only its own preview

The handler blanked fileMetadata and filePreview without checking that its
request still owned the pane, so selecting a second file while the first was
still loading meant the first's rejection wiped the second's preview.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* fix: key preview ownership on the request, not the selected key

Comparing the selected key let an older request speak for a newer one when both
targeted the same key, which a storage switch does, and made a request whose
selection had moved to something with no metadata return early with the spinner
still up — the case the handler exists to prevent.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* fix: clear the preview when the previewed file is deleted

The lazy branch refetches only the affected level and returns, so it never
reached the reset that the flat refresh gets from clearAndLoadFiles. The pane
renders from fileMetadata rather than from the selection, leaving the deleted
file previewed with working download, move and delete actions.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* fix: retire the in-flight preview load when its file is deleted

Clearing the pane was not enough: a metadata response computed before the DELETE
landed still repopulated it, restoring the deleted file's preview and its
download, move and delete actions. Deleting now retires the owning request, and
the success and preview writes honour that the same way the failure path does.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* fix: clear the preview loading flag when the delete retires its request

Retiring the in-flight metadata load left nobody to report its outcome, so in
lazy mode the pane sat on "Loading..." instead of falling back to the empty
state. The delete owns the flag once it has retired the request.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* chore: drop the regenerated openapi deref artifacts

They are generated files that CI only syntax-validates, never checks against
openapi.yaml, and the committed copies already differ from the spec they derive
from by ~9.7k lines. Regenerating here imported that pre-existing drift into a
feature diff, burying ~800 lines of actual change under ~17k lines of other
changes' staleness. Regenerating them is its own chore.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* docs: state the flat cursor invariant once, where the cursor lives

It was spelled out at four sites, which is what AGENTS.md asks not to do. The
rule now sits on the declaration it constrains and the guards reference it.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* nit ui

* fix: add the paged listing to the served openapi json

openapi_json() embeds openapi-deref.json via include_str!, and the Docker build
regenerates only the yaml artifact, so the json is served exactly as committed —
leaving the new operation out of the Scalar API reference.

Spliced in the operation and the two schemas it references rather than
regenerating, which would have re-imported ~7k lines of pre-existing drift
between the committed artifact and the spec it derives from.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* chore: bump ee-repo-ref for the filesystem symlink boundary

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* chore: update ee-repo-ref to 0373b4bfdaf8dd51533552e2e4de63ceb3c18b4d

This commit updates the EE repository reference after PR #697 was merged in windmill-ee-private.

Previous ee-repo-ref: eb1a765bb9b29e0c94a6e4942c304934fa15406e

New ee-repo-ref: 0373b4bfdaf8dd51533552e2e4de63ceb3c18b4d

Automated by sync-ee-ref workflow.

---------

Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
Co-authored-by: windmill-internal-app[bot] <windmill-internal-app[bot]@users.noreply.github.com>
Co-authored-by: Ruben Fiszel <ruben@windmill.dev>
2026-08-04 01:41:06 +02:00
GuilhemandClaude Opus 4.8 c5060a1e9a fix: scope AI-session flow/script editors to the session workspace (#10025)
* fix: scope flow script-edit drawer to session workspace and fix scroll

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

* fix: scope flow schema inference to session workspace

Thread an optional workspace through loadSchemaFromPath/loadSchemaFlow/
loadSchemaFromModule/loadFlowModuleState/initFlowState/pickScript/pickFlow
and pass the op (session) workspace at fork-context call sites, so a flow
opened in an AI session resolves path-referenced scripts/subflows against
the session workspace instead of the nav workspace.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

* fix: scope script editor log panel and git-repo pickers to op workspace

LogPanel and the ansible git-repo viewer/picker read the nav workspace
directly; pass the script editor's op workspace so past-test results/logs
and git-repo resource/file lookups target the session workspace in a fork.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

* docs: correct session pipeline trigger-editor workspace comment

The comment claimed session activation syncs $workspaceStore; SessionPicker
intentionally does not, so trigger create/edit/delete from a fork session's
pipeline canvas writes to the nav workspace. Document the known limitation.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

* fix: forward op workspace to git-repo S3 file browser

GitRepoViewer scoped its own calls to the op workspace but rendered the
nested S3FilePickerInner without workspace={ws}, so the file list/preview/
metadata still queried the nav workspace with a session-workspace prefix.
Addresses Codex review on #10025.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

---------

Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-07-10 13:51:16 +00:00
Ruben FiszelandClaude Opus 4.7 a03f5c0fab [ee] fix: GitRepoViewer reliable load for large repos (#8991)
Fixes silent timeouts and partial-tree rendering when loading large git
repositories into the in-app viewer (Tony Hoang report: 400+ host_vars
files, 32 roles).

Three coordinated fixes:

1. Frontend (GitRepoViewer.svelte): drop the 60s clone timeout. Long-poll
   getJobUpdates until the job completes, with a 30 min hard cap and a
   user-cancel button. Stream live job logs into the viewer with a link
   to the full job page. After success, verify the
   .windmill_clone_complete marker before flipping pathExists, so a
   partial S3 directory is no longer rendered as a complete tree.

2. Backend (check_s3_folder_exists, EE): new optional marker_file query
   param. When set, the handler short-circuits to head() on the marker
   object instead of "any object under prefix exists". The frontend now
   always passes marker_file=.windmill_clone_complete.

3. Hub script: cloneRepoToS3forGitRepoViewer points at hub/28216, which
   uploads files via a bounded-concurrency pool (16 workers), emits
   throttled progress logs, and writes .windmill_clone_complete as its
   last action. docs/clone_repo_and_upload_to_instance_storage.bun.ts is
   the source for that hub publish; docs/git-repo-viewer-hub-script.md
   explains the change.

Also drops three unused legacy hubPaths entries
(cloneRepoToS3forGitRepoViewer_0..2) — none were referenced from
anywhere in the codebase.

Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-04-30 13:28:50 +00:00
Ruben FiszelandClaude Opus 4.7 ddf14a51d8 refactor: split git repo viewer effects to remove redundant calls (#8943)
* refactor: split git repo viewer effects to remove redundant calls

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

* refactor: compact file preview header in s3 file picker

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

* refactor: skip empty preview status row when no message applies

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

---------

Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-04-27 04:46:22 +00:00
Ruben FiszelandClaude Opus 4.7 b8fcb7f04b fix: preserve s3 rootPath when reloading git repo viewer (#8942)
Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-04-27 04:23:19 +00:00
wendrul e719be54e0 Fix ansible delegate mode only working with ssh (#6858) 2025-10-17 13:58:00 +00:00
wendrul 83337b2035 Ansible repo update hubPath + fixes (#6843)
* Allow ssh auth

* ssh changes pt2

* Fix unused

* Update package.json and hubPaths
2025-10-16 18:24:41 +00:00
wendrulandRuben Fiszel 32fae7a10c feat: ansible playbook execution git repo mode (repo viewer + UI utils) (#6831)
* Improve minio on flake.nix

* Add first asset parsing logic for ansible

* Correct html gt sign

* Decouple s3 file picker from drawer

* Factor duplicate code into snippet

* Update S3FilePickerInner to be compatible

* Fix pane shrinking issue

* Git repo viewer

* Change GitRepoViewer

* Endpoints for git repo visualizer

* Move git repo viewer to its own component

* Add button to populate git repo viewer

* Update parser yaml for new ansisble features (repo viewer)

* Reflect parser changes for ansible

* Add button to add the git repo mode of declaration for ansible

* Factor function

* Playbook + inventories into the drawer

* Add button to add inventories from s3

* Move tests to lib.rs

* Inventory loading from s3

* Move get github app token logic to be reused by ansible

* Update parser and ansible executor

* Use the correct path for inventories

* Add nushell to flake for wasm builds

* Add published parser

* Update hubPaths with clone and upload to s3

* Update ee-repo to the branch ref

* Fix npm run check

* Update cargo.lock

* Change labels on buttons

* Remove debug log

* Update ee-repo-ref

* Fix ee issues

* Update ee-repo ref

* Fix typo

* Fix ee

* Update ee-repo-ref

* Fix missing imports

* Unused var

* Fix typo

* Layout improvents

* Fix typos

* Remove unused function and log

---------

Co-authored-by: Ruben Fiszel <ruben@windmill.dev>
2025-10-16 09:08:18 +00:00