AutoComplete

AutoComplete allows you to suggest values to the user while they are typing in a form field. Instead of filling in the full value manually, the system queries an external service and displays relevant suggestions (e.g. city names, street names, product codes).

This feature improves the user experience by making forms faster to fill out and reducing input errors.

How AutoComplete works

1. The user starts typing into a field, for example City.
2. PalDock runs the scenario attached to that field, which calls an external service. See Connection Creator in Structures.
3. The service responds with a list of possible values.
4. The suggestions are displayed to the user as they type.
5. When the user picks one, the value is filled into the field.

The user can still type a value that was not suggested. AutoComplete offers help, it does not restrict what the field accepts. If the value has to be one of a known set, use a Select field or add validation.

Setting it up

Three things have to line up:

• The scenario calls the service and passes in what the user has typed so far.
• The list of suggestions is picked out of the response and stored under a name, for example names.
• The field is linked to the scenario, and its accessor points at that name, so PalDock knows which part of the response holds the suggestions.

If the accessor points at something that is not a list, no suggestions appear and nothing is reported. That is the first thing to check when a field stays silent.

Example: city names from OpenStreetMap

The field in the structure has the system name city.

The scenario sends a GET request to https://nominatim.openstreetmap.org/search with:

• the value the user is typing, mapped to the city parameter
• featureType=city
• format=json
• limit=5
• a User-Agent header, which this particular service requires

The response is a list of places. The scenario picks the name out of each one and stores the result as names. On the field, the accessor is set to names.

Note the limit=5. A suggestion list is read at a glance, so a long one is worse than a short one, and every extra result costs response time the user is waiting through.

Typical use cases

• Suggesting city or street names from geolocation services.
• Offering product codes or SKUs from a product feed.
• Completing company names from a business registry.
• Providing postal codes based on partial input.

What to watch out for

⚠️ AutoComplete depends on external services. If the service is unavailable or slow to respond, suggestions do not load.

• Keep it fast. The scenario runs while the user is mid-word, so the whole thing has to finish in a couple of seconds to be useful at all. Set a short timeout on the HTTP node. See Limits and timeouts.
• An outage is not a blocker here. Unlike validation, a failed AutoComplete does not stop the form. The user sees no suggestions and types the value themselves. That is a good reason to use AutoComplete rather than a Select when the list is long and the service is not fully reliable.
• Watch the call volume. The service is queried while the user types, not once on submit, so one filled form is several requests. Check what your provider allows and what it costs before you put AutoComplete on a high-traffic form.