Summary
This month, we’re less focused on shipping new features and more focused on polishing what we have for production. We’re customer-driven, making improvements for user experience and focused on shipping higher value products.
TAP
With TAP in production and the planned transition to indexer-rs announced, we really need to make sure that we focus on ensuring TAP v1 is stable and bug-free. Bug fixes, documentation updates, and refactors have been our highest priority. We’ve seen good adoption already, with almost 60% of the queries on the network going through TAP.
Easing Indexers’ migrations and supporting bugfixes will be our top priority in the next month. At the same time, we’re hard at work wrapping up TAP v1. We hope to soon shift focus to TAP v2, which will enable TAP to be used for multiple data services.
Features and Fixes
- Especially important for those of you who are migrating to
indexer-rsand TAP, we created a migration configuration that injects environment variables from the old stack into the new stack, making it easier to switch. - tap-agent is now more reliable while closing allocations against malfunctioning subgraphs.
- We improved our tracker for unaggregated fees, and we expect it to be much more reliable, especially with high-volume indexers
- We modified the code to fail early at startup if components like subgraphs are not working correctly, rather than waiting for things to crash during the code’s runtime.
- We pushed a couple of performance improvements
- Concurrent component creation, resulting in faster start-up
- RAVs with concurrent requests.
Newest Versions
indexer-service-rs: 1.2.1indexer-tap-agent: 1.4.1indexer-agent: 0.21.6
Verifiable Extraction
Pre-merge Ethereum Verifiable Extraction (VE) is in a transition phase from research code to production code. We’re in touch with several people who want to integrate VE into their stacks. Maintaining a tight feedback loop with our users is key to ensuring that our verification algorithm is made a part of software that users actually wish to use. Next month is going to be a continuation of this work.
Speaking of those VE users, we’ve also understood a bit more about how they want to run things. Our previous solution required users to run some infrastructure and tie multiple pieces together manually. This adds a lot of complexity and effort for VE users. As a result, we’re abandoning our post-merge approach from last month for a new approach. This will be our focus over the next month. This is still in the research stage, but it should make post-merge and pre-merge VE look identical from a user’s perspective. Right now, our focus is on validating the new path post-merge with a PoC.
AI
We have released our paper on synthetic data generation, which can be found here. Currently, we are in the process of conducting user research around delegation and applications of natural language as an interface to The Graph. We encourage people to participate in these endeavors, and you can sign up through the following forms (delegation | natural language). We are excited about the progress on these two fronts and hope to push towards active development soon.