76 results found
-
Postgres Functions exposed as an API
Similar services, like postgraphile, allow exposing queries / mutations as postgres functions. It's extremely powerful, and allow extremely custom queries / logic for exposing data to the clients.
13 votes -
Table Event Triggers
In additional to the
onMutationExecuted, we would love to have events on table mutations.onRowUpdatedonRowCreatedonRowDeleted
onMutationExecutedis already launched in Nov 2026. https://firebase.blog/posts/2025/11/dataconnect-nov2512 votes -
Evaluate Mutation Results in Realtime Refresh Conditions & Support Mutation-Side Triggers
Currently, when configuring a refresh condition on a query, the condition can only evaluate the triggering mutation's variables and auth (uid/token). We cannot check the actual result (response) of the mutation.
This creates a significant friction point when mutations are executed from non-client environments, such as a Cloud Function triggered by another backend service (e.g., Firebase Storage). Because the Cloud Function executes without the end-user's auth context, the only way to accurately filter the query refresh is to pass the userId into the mutation's variables. This feels like an unnecessary workaround just to satisfy the refresh condition.
To make real-time…
11 votes -
AlloyDB Datasource
Support AlloyDB as a datasource for Data Connect in addition to Cloud SQL via the same PostgreSQL bindings
10 votes -
Advanced Vector Embeddings with pgvector
Create and store advanced embeddings for Data Connect types in pgvector through support of configurable chunking strategies and automatic embedding by annotations.
9 votes -
Custom GraphQL Scalar Support
On top of common scalars such as Email, Phone, etc., custom scalars are defined by developers on the client-side application, with logic for marshaling and unmarshalling the data provided, which could be either a class with the two methods or a couple of functions.
8 votes -
Data Transformation With CEL expression
I love that data connect supports
{.._expr: "auth.uid"}to define server value based on Firebase auth token.What about other CEL operations?
I want the ability to perform simple transformation using GraphQL variables, Firebase auth token and current timestamp. For example:
_expr: "'hello from ' + auth.email"I want the ability to combine data from different sources. Like execute a query and then use its result to perform another query or mutation. CEL expression can help cross reference data in previous steps.
8 votes -
Vertex AI Integration
Add deeper integration with Vertex AI to Data Connect, for example allowing direct LLM content generation and simple RAG use cases.
8 votes -
Enhanced Foreign Key Behavior
I am reaching out to share a suggestion regarding the current implementation of foreign key constraints in Firebase DataConnect.
Currently, it appears that the default behavior for foreign keys is automatically assigned during schema deployment, and there is no way to explicitly define ON DELETE and ON UPDATE rules (e.g., NO ACTION, CASCADE) in the GraphQL schema. This limitation has the following implications:
1. Deployment Overrides: Even if these rules are manually adjusted in the underlying PostgreSQL database, deploying an updated schema resets the rules to the platform’s defaults.
2. Lack of Customization in Schema: Adding support for directives like…7 votes -
7 votes
-
Sending arrays as parameters, to allow scaling.
When building simple select queries in GraphQL (really simple ones), the underlying query is being transformed, sending the values through as parameters.
As you can use arrays as parameters, that would allow the DBs to scale as it makes this more robust. Without scaling, it will limit production applications form fully investing.
As an example you run into instances where the select for 50 columns in the schema (a standard employment record, nothing special) is transformed and we end up getting the dreaded 100 parameters error. This is a Postgres issue, but can be avoided.
The Error for the fun…
5 votes -
Using Genkit embedders (including from e.g. genkitx-openai-plugin)
Instead of only being able to select embedding models from Vertex AI (e.g. Gecko), it'd be great to be able to pass in any embedder imported from Genkit plugins like genkitx-openai-plugin or genkitx-cohere
5 votes -
Cloud Spanner Datasource
Support Cloud Spanner as a datasource for Data Connect in addition to Cloud SQL via Cloud Spanner's PostgreSQL compatibility layer.
5 votes -
Map Firebase Data Connect (FDC) schema to custom postgres schema (not PUBLIC)
Currently, any gql files placed in schema folder generate one FDC schema in FDC console, which generates associated tables in postgres PUBLIC schema. Putting tables there is not ideal.
It would be nice if creating additional subfolders within schema source folder (or some other mechanism) could allow mapping to developer defined postgres schemas other than PUBLIC. For example, the following folder structure:
schema/appname/*.gql
would map to APPNAME schema in Postgres and tables would be created there.
4 votes -
Aggregation Order by Value
With data connect aggregation queries, I would like to order the returned results by an aggregated value. The below example fails with the message: "Field "count" is not defined by type "LeadOrder". Did you mean ...?"
query {
leads(orderBy: [{ _count: DESC }], limit: 10) {
company
_count
}
}4 votes -
Add service to Data Connect Triggers
Trigger a Cloud Function that will add the service and schema to Data Connect when the user signs up successfully.
4 votes -
SQL Connect should support more postgreSQL constraints and indexes
Customers that have an existing schema with PostgreSQL or have product requirements that involve using more complex PostgreSQL features like CHECK constraints or partial indexes cannot use fdc because it tries to drop the schema and fails to start up.
There's partial support for this already via the brownfield setup guide but if constraints and indexes could be included this would unlock more use cases. https://firebase.google.com/docs/sql-connect/manage-services-and-databases#integrate-existing-databases
See the issue here https://github.com/firebase/firebase-tools/issues/10571
3 votes -
More information in errors for native sql statements
Errors in native sql statements like in _execute always result in the same "Invalid SQL statement" error without any further information, which makes it really difficult to debug. More data - for example the part of the SQL statement that is invalid - would be really helpful! Especially as this error also seems to be thrown when the SQL is perfectly fine but a wrong _execute syntax is used, which should not throw an "Invalid SQL statement" error, as it is not about SQL!
3 votes -
Support Multiple Data Connect Services in Emulator Suite (or Allow Hybrid Cloud Fallback)
Currently, the Firebase Emulator Suite allows running only a single Data Connect service instance locally. For projects utilizing a microservices architecture with multiple isolated Data Connect services, this creates a significant blocking issue during development.
Primary Feature Request: Multi-Service Emulation - Please update the Emulator Suite to support running multiple Data Connect services simultaneously.
Alternative / Interim Solution: Hybrid Cloud Fallback - If full multi-service emulation is technically complex to implement immediately, please introduce a "Hybrid" mode as an alternative.
Current Behavior: When the emulator is running, all calls to Data Connect are forced to look locally. If a specific…
3 votes -
SolidJS SDK for data connect
I need to have this SDK because for now only React and Angular have SDK.
3 votes
- Don't see your idea?