Change the behaviour where not passing attribs to the update API
overwrites the attribs with empty values. This commit changes the
behaviour so that in the absence of the attribs field in the
subscriber API, the existing value in the DB is retained.
If a user is already subscribed to an optin list but hasn't
confirmed, subscribing using the same e-mail id from the public
form now re-sends the optin e-mail while also showing an
appropriate message on the frontend rather than just saying
"subscribed successfully".
https://github.com/knadh/listmonk/issues/266https://github.com/knadh/listmonk/issues/264
In addition to generating HTML forms for selected public lists,
the form page now shows a URL (/subscription/form) that can be
publicly shared to solicit subscriptions. The page lists all
public lists in the database. This page can be disabled on the
Settings UI.
This commit removes the Go html2text lib that would automatically
convert all HTML messages to plaintext and add them as the alt
text body to outgoing e-mails. This lib also had memory leak
issues with certain kinds of HTML templates.
A new UI field for optionally adding an alt plaintext body to
a campaign is added. On enabling, it converts the HTML message in
the campaign editor into plaintext (using the textversionjs lib).
This introduces breaking changes in the campaigns table schema,
model, and template compilation.
Lists, campaigns, and subscribers tables now support server-side
sorting from the UI. This significantly changes the internal
queries from prepared to string interpolated to support dynamic
sort params.
The link_clicks.link_id table was NULLable incorrectly. Links that
do not exist should not register a tracking entry. Fix the query
and also update the schema + migration (breaking table change).
A new toggle switch in Settings -> Privacy, which is off by
default, allows campaign views (pixel) and link clicks to function
without registering the subscriber ID against view and click
events, anonymising tracking. When off, the subscriber UUIDs in
view and link tracking URLs are removed, anonymising subscriber
information from HTTP logs as well.
This is a major feature that builds upon the `Messenger` interface
that has been in listmonk since its inception (with SMTP as the only
messenger). This commit introduces a new Messenger implementation, an
HTTP "postback", that can post campaign messages as a standard JSON
payload to arbitrary HTTP servers. These servers can in turn push them
to FCM, SMS, or any or any such upstream, enabling listmonk to be a
generic campaign messenger for any type of communication, not just
e-mails.
Postback HTTP endpoints can be defined in settings and they can be
selected on campaigns.
installations.
In v0.7.0's automatic DB migration, the default value 'enabled'
of subscribers.status was missed incorrectly, causing imports to
break on auto-upgraded versions. This is now fixed by explictly
inserting the status field in CSV imports.
- Add missing `app.root_url` key in migration.
- Register `/settings` handler in the backend.
- Add dummy dots in secret fields on the UI for visibility.
This is a major breaking change that moves away from having the
entire app configuration in external TOML files to settings being
in the database with a UI to update them dynamically.
The app loads all config into memory (app settings, SMTP conf)
on boot. "Hot" replacing them is complex and it's a fair tradeoff
to instead just restart the application as it is practically
instant.
A new `settings` table stores arbitrary string keys with a JSONB
value field which happens to support arbitrary types. After every
settings update, the app gracefully releases all resources
(HTTP server, DB pool, SMTP pool etc.) and restarts itself,
occupying the same PID. If there are any running campaigns, the
auto-restart doesn't happen and the user is prompted to invoke
it manually with a one-click button once all running campaigns
have been paused.
- Fix path related issues in filesystem and S3.
- Add checks for S3 "/" path prefix.
- Add support for custom S3 domain names.
- Remove obsolete `width` and `height` columns from media table (breaking)
- Add `provider` field to media table (breaking)
- antd+react was resulting in extremely clunky and unreadable
spaghetti frontend code (primarily due to how antd is).
- Buefy is lighter by an order of magnitude, has excellent
responsive views (especially tables) and usability.
- Vue's templating produces far more readable template code.
- Campaigns now have a `type` property (regular, opt-in)
- Opt-in campaigns work for double opt-in lists and e-mail
subscribers who haven't confirmed their subscriptions.
- Lists UI shows a 'Send opt-in campaign' optin that
automatically creates an opt-in campaign for the list
with a default message body that can be tweaked before
sending the campaign.
- Primary usecase is to send opt-in campaigns to subscribers
who are added via bulk import.
This is a breaking change. Adds a new Postgres enum type
`campaign_type` and a new column `type` to the campaigns table.
- Lists can now be marked as single | double optin.
- Insert subscribers to double opt-in lists send out a
confirmation e-mail to the subscriber with a confirmation link.
- Add `{{ OptinURL }}` to template functions.
This is a breaking change. Adds a new field 'optin' to the lists
table and changes how campaigns behave. Campaigns on double opt-in
lists exclude subscribers who haven't explicitly confirmed subscriptions.
Changes the structure and behaviour of how notification e-mail routines,
including notif email template compilation, notification callbacks for
campaign and bulk import completions.
- Toggle options to enable self-service data export and wipe
options on the public unsubscription page. Subscribers can get
a copy of all data on them e-mailed to them as JSON, or
instantly wipe all their data.
- Refactor "unsubscribe" pages and URIs to "subscription".
- Add export icon to subscriber admin view.
- Simplify campaigns querying to separate statistics gather into
a separate query for lazy loading.
- Simplify subscribers query to separate list fetching into
a separate query for lazy loading.
The get-campaigns query was doing two direct joins with the campaign_views
and link_clicks tables (that have very large number of relationships)
to get the view and click counts. Now the campaigns are selected first
in a CTE and their views and counts are aggregated in two more CTEs,
and the whole thing is then aggregated to produce the final results.
- Add a name / e-mail "quicksearch" input to the UI
- Implement row selection and aggregation at table level and a "select all" that selects all rows at the query level
- On selected subscribers, add bulk list management (add / remove / unsubscribe), blacklist, and delete
Kailash Nadh
thejaswinipathi