- Convert `nodeHelpers` mixin into composable and fix types
- Replace usage of the mixin with the new composable
- Add missing store imports in components that were dependent on opaque
imports from nodeHelpers mixin
- Refactor the `CollectionParameter` component to the modern script
setup syntax
Github issue / Community forum post (link here to close automatically):
---------
Signed-off-by: Oleg Ivaniv <me@olegivaniv.com>
## Summary
Prevent opening NDV search if `/` is typed in a contenteditable element
...
#### How to test the change:
1. Create a workflow with a Code node that has some input and output
2. Run the workflow and then open the Code node
3. Type `/` anywhere in the Code node
#### Expected behavior:
NDV search should not be opened and focused
This PR introduces the possibility of inviting new users with an `admin`
role and changing the role of already invited users.
Also using scoped permission checks where applicable instead of using
user role checks.
---------
Co-authored-by: Val <68596159+valya@users.noreply.github.com>
Co-authored-by: Alex Grozav <alex@grozav.com>
Co-authored-by: Iván Ovejero <ivov.src@gmail.com>
- Split Items List node into separate nodes per action
- Review node descriptions
- New icons
- New sections in subcategories
---------
Co-authored-by: Giulio Andreini <andreini@netseven.it>
Co-authored-by: Deborah <deborah@starfallprojects.co.uk>
Co-authored-by: Michael Kret <michael.k@radency.com>
This PR should improve performance when opening large workflows that
contain node issues (Fixes ADO-1521)
**Background**
Main reason for this poor performance is that our `getCurrentWorkflow()`
store getter is unnecessarily heavy but on top of that we are calling it
more than we need. This addresses the second part of the issue by
changing the following:
- Pausing node issue processing while workflows are loading
- Only getting current workflow once (instead for every node) when
calling `refreshNodeIssues`
**Benchmark**
This was tested on a workflow attached to [this Linear
ticket](https://linear.app/n8n/issue/ADO-1501/deliveryhero-enterprise-instance-very-slow-loading-workflows)
and this fix brings down opening time from **~1m10s** to **~28s** on my
laptop.
**Tests**
- [Latest e2e tests
run](https://github.com/n8n-io/n8n/actions/runs/7060874994)
https://community.n8n.io/t/ui-very-slow-with-more-than-100-nodes/8236/14
This PR sets the stage for the node creator to handle sections within
subcategories. No visible changes result from this PR; the next PR will
define sections and assign nodes accordingly.
Sections are configurable in
`packages/editor-ui/src/components/Node/NodeCreator/viewsData.ts`:
```
{
type: 'subcategory',
key: FILES_SUBCATEGORY,
category: CORE_NODES_CATEGORY,
properties: {
title: FILES_SUBCATEGORY,
icon: 'file-alt',
sections: [
{
key: 'popular',
title: i18n.baseText('nodeCreator.sectionNames.popular'),
items: ['n8n-nodes-base.readBinaryFiles', 'n8n-nodes-base.compression'],
},
],
},
},
```
For example:
<img width="302" alt="image"
src="https://github.com/n8n-io/n8n/assets/8850410/74470c07-f4ea-4306-bd4a-8d33bd769b86">
---------
Co-authored-by: Michael Kret <michael.k@radency.com>
The template credentials were matched before solely based on the
credential name on the template. This fixes the matching to use both
credential type name and credential name.
Removing scope permission checks on workflow and credential sharing and
relying only on resource ownership.
Every user can share only the workflows and credentials they created.
Moving cloud hooks and store initialization logic after users are
authenticated. This will ensure user local account is available when
their cloud plan data is being fetched.
This PR also adds the following error handling improvements:
- Added error handling to the same initialization logic
- Fixed empty `catch` clauses inside the cloud store which caused it to
silently fail and complicated debugging of this bug
This PR fixes mouse position detection in workflow previews (executions
preview and workflow history view):
https://github.com/n8n-io/n8n/assets/2598782/83237b76-b099-4560-9c47-a963f0848297
**Cause**:
We were setting `isDemo` state in canvas store only when the `NodeView`
component is activated, which never happens for `NodeView` instances
that are kept alive (above mentioned workflow previews)
**Fix**:
This workflow updates store state also when the `NodeView` component is
mounted.
extracted out of #7336
---------
Co-authored-by: Jan Oberhauser <jan.oberhauser@gmail.com>
Co-authored-by: Oleg Ivaniv <me@olegivaniv.com>
Co-authored-by: Alex Grozav <alex@grozav.com>
This PR introduces the following changes:
- New Vue stores: `collaborationStore` and `pushConnectionStore`
- Front-end push connection handling overhaul: Keep only a singe
connection open and handle it from the new store
- Add user avatars in the editor header when there are multiple users
working on the same workflow
- Sending a heartbeat event to back-end service periodically to confirm
user is still active
- Back-end overhauls (authored by @tomi):
- Implementing a cleanup procedure that removes inactive users
- Refactoring collaboration service current implementation
---------
Co-authored-by: Tomi Turtiainen <10324676+tomi@users.noreply.github.com>
Followup to #7566 | Story: https://linear.app/n8n/issue/PAY-926
### Manual workflow activation and deactivation
In a multi-main scenario, if the user manually activates or deactivates
a workflow, the process (whether leader or follower) that handles the
PATCH request and updates its internal state should send a message into
the command channel, so that all other main processes update their
internal state accordingly:
- Add to `ActiveWorkflows` if activating
- Remove from `ActiveWorkflows` if deactivating
- Remove and re-add to `ActiveWorkflows` if the update did not change
activation status.
After updating their internal state, if activating or deactivating, the
recipient main processes should push a message to all connected
frontends so that these can update their stores and so reflect the value
in the UI.
### Workflow activation errors
On failure to activate a workflow, the main instance should record the
error in Redis - main instances should always pull activation errors
from Redis in a multi-main scenario.
### Leadership change
On leadership change...
- The old leader should stop pruning and the new leader should start
pruning.
- The old leader should remove trigger- and poller-based workflows and
the new leader should add them.
1. Reduce a lot of code duplication
2. Move more endpoints out of `Server.ts`
3. Move all query-param parsing and validation into a middleware to make
the route handlers simpler.
This PR:
- Creates `InvitationController`
- Moves `POST /users` to `POST /invitations` and move related test to
`invitations.api.tests`
- Moves `POST /users/:id` to `POST /invitations/:id/accept` and move
related test to `invitations.api.tests`
- Adjusts FE to use new endpoints
- Moves all the invitation logic to the `UserService`
---------
Co-authored-by: कारतोफ्फेलस्क्रिप्ट™ <aditya@netroy.in>
Github issue / Community forum post (link here to close automatically):
---------
Co-authored-by: Giulio Andreini <g.andreini@gmail.com>
Co-authored-by: कारतोफ्फेलस्क्रिप्ट™ <aditya@netroy.in>
This change expands on the command channel communication introduced
lately between the main instance(s) and the workers. The frontend gets a
new menu entry "Workers" which will, when opened, trigger a regular call
to getStatus from the workers. The workers then respond via their
response channel to the backend, which then pushes the status to the
frontend.
This introduces the use of ChartJS for metrics.
This feature is still in MVP state and thus disabled by default for the
moment.
Github issue / Community forum post (link here to close automatically):
---------
Co-authored-by: Giulio Andreini <g.andreini@gmail.com>
Co-authored-by: Marcus <marcus@n8n.io>
useStorage takes the default value `undefined` and sets it in local
storage.. also returns "undefined" as string which breaks onboarding
flows
Github issue / Community forum post (link here to close automatically):
Because of recent changes does the frontend break when a node is not
known.
Github issue / Community forum post (link here to close automatically):
---------
Signed-off-by: Oleg Ivaniv <me@olegivaniv.com>
Co-authored-by: Oleg Ivaniv <me@olegivaniv.com>
Github issue / Community forum post (link here to close automatically):
---------
Co-authored-by: कारतोफ्फेलस्क्रिप्ट™ <netroy@users.noreply.github.com>
Currently Trigger-Nodes and Sub-Nodes display options that do not make
sense or work like:
- Always Output Data
- Execute Once
- On Error
This PR hides those options for nodes that can not make use of them
anyway to not confuse users.
Github issue / Community forum post (link here to close automatically):
Allow overriding theme through query params.. to be able to override it
from preview iframe in webcomponent
Github issue / Community forum post (link here to close automatically):
Github issue / Community forum post (link here to close automatically):
---------
Co-authored-by: कारतोफ्फेलस्क्रिप्ट™ <aditya@netroy.in>
Co-authored-by: Giulio Andreini <andreini@netseven.it>
Fixes the issue that currently no Switch-Node can be created by for
example pressing + on the parent node or via tab when another node is
already selected.
Github issue / Community forum post (link here to close automatically):
---------
Signed-off-by: Oleg Ivaniv <me@olegivaniv.com>
Co-authored-by: Oleg Ivaniv <me@olegivaniv.com>
extracted out of #7336
---------
Co-authored-by: Jan Oberhauser <jan.oberhauser@gmail.com>
Co-authored-by: OlegIvaniv <me@olegivaniv.com>
Co-authored-by: Jan Oberhauser <janober@users.noreply.github.com>
Co-authored-by: Val <68596159+valya@users.noreply.github.com>
Co-authored-by: Alex Grozav <alex@grozav.com>
Co-authored-by: Deborah <deborah@starfallprojects.co.uk>
Co-authored-by: Jesper Bylund <mail@jesperbylund.com>
Co-authored-by: Jon <jonathan.bennetts@gmail.com>
Sometimes canvas selection stops working after users interact with node
action buttons (for example if node is moved by dragging one of the
buttons)
NOTE: Ticket number in the branch name is wrong, this fixes ADO-1226
## Issue
In community edition, clicking on "View plans" button on "Settings" ->
"Usage and plan" page (e.g. http://127.0.0.1:5678/settings/usage) opens
two new tabs with n8n pricing (one of them with UTM tracking, another
without).
This was introduced in #6317 , when click handler of "View plans" link
container [started
calling](https://github.com/n8n-io/n8n/pull/6317/files#diff-0bf26afac8a06e03b3d39d0668f22408859355b585a9ab420800c125e33f0691R109)
`uiStore.goToUpgrade(...)` which opens n8n pricing in a new tab, while
browser opens another tab for the link URL.
The simplest fix, implemented in this PR, is to prevent default event
handling (so that, after `onViewPlans` is called, browser will not
attempt to process the click additionally as clicking on the link),
similarly to how it is prevented on some other pages. It only solves the
immediate problem of browser opening two new tabs on clicking "View
plans".
Note that **I didn't implement any tests for the changed behavior**,
because it was not covered by tests before, and I couldn't quite figure
out how to cover it now within the existing test approach (considering
that testing the fact that only one new tab is open will likely require
to write entirely new tests relying on puppeteer; as far as I can see,
no existing `editor-ui` tests are doing anything like that). I'll gladly
implement tests for the new behavior if you tell me how you would like
them to look.
The existing tests for `editor-ui` still pass; I didn't run tests for
other subpackages (see "additional contribution notes" below).
## Additional notes on the issue.
I'm not sure that the change in this PR is the correct long-term
solution for the issue, because the URLs for these two methods (custom
click handler for link container and default link handling) are slightly
different:
* Custom click handler calls `useTelemetryStore().track('User clicked
upgrade CTA', ...)`; then calls `sendUsageTelemetry('view_plans')` (it
feels weird that two calls to telemetry are made); then opens new tab
for `https://n8n.io/pricing?utm_campaign=open&source=usage_page` (note
that prior to #7316 the second call to telemetry was done after the new
tab is opened, not before);
* Link itself refers to another page, with slightly different tracking
parameters:
`https://subscription.n8n.io/?instanceid=[REDACTED]&version=1.10.0&callback=http%3A%2F%2F127.0.0.1%3A5678%2Fsettings%2Fusage&source=usage_page`;
but this page redirects to `https://n8n.io/pricing/`.
It is not clear which one of the two is the right way of doing things.
Although `goToUpgrade` is called in 20 places throughout `editor-ui`,
while `viewPlansUrl`, as far as I can see, is used for this button only.
Additionally, since Settings pages don't work without JS anyway, I can
only think of two separate scenarios where any tab would be opened:
* Left-clicking the link (or Ctrl-clicking, or pressing Space or Enter
when the link is focused, or tapping): previously, both custom click
handler was executed and link's `href` was opened; in this PR, only
custom click handler is executed (similarly to how it is done in the
other places where `goToUpgrade` is called);
* Right-clicking (or long tapping, or opening context menu in any other
way) and selecting "open link in new tab" (or similar): opens a new tab
for URL from the `href` attribute (and does not send any telemetry at
all).
I'd say that the better permanent solution would probably be to get rid
of one of these methods entirely, and only rely on another in all cases
(for me, as an outside contributor, the preferred way would be for
custom click handler to only send telemetry, while letting my browser
handle the actual navigation). However, that would be a large change,
much more than one line in this PR.
Additionally, other similar places where `goToUpgrade` is currently
called (directly or indirectly) would also need to be adapted for this
change.
## Additional contribution notes
As a first-time contributor, I've encountered several things I didn't
expect; I'm not sure if they should be expected or are issues:
1. Tests for the entire monorepo consume a lot of RAM; 20GB free RAM was
not enough, so I couldn't run tests for the entire monorepo and had to
only run them for `packages/editor-ui`;
2. Linting is very slow; `pnpm lint` in `packages/editor-ui` takes ten
minutes to complete;
3. It seems that types are not checked. Code OSS highlights numerous
errors in code files: for example, `'debug'` is incompatible with
`CloudUpdateLinkSourceType` expected by `goToUpgrade` here:
3e7a4d3b2c/packages/editor-ui/src/composables/useExecutionDebugging.ts (L128)
However, I'm not getting any errors during build. There is a `typecheck`
script defined in `package.json`, but `pnpm typecheck` fails with:
```
n8n-toy-demo:~/projects/n8n/packages/editor-ui$ pnpm typecheck
> n8n-editor-ui@1.10.0 typecheck
/home/inga/projects/n8n/packages/editor-ui
> vue-tsc --emitDeclarationOnly
error TS5069: Option 'emitDeclarationOnly' cannot be specified without
specifying
option 'declaration' or option 'composite'.
Found 1 error.
ELIFECYCLE Command failed with exit code 1.
n8n-toy-demo:~/projects/n8n/packages/editor-ui$
```
Replacing `--emitDeclarationsOnly` with `--noEmit` in `package.json`
unblocks typechecking and results in seemingly, at first glance, correct
"Found 1924 errors in 306 files" (at least several of the reported
errors that I've checked seem to be correct).
But maybe I'm missing something and there are not in fact two thousands
type errors in `editor-ui`?
- Fix pinch-to-zoom
- Support command + scroll to zoom
- Improve accuracy of zooming (scroll more = zoom more)
- Zoom limits
- Zoom relative to mouse position
