Hi Cole, Thank you for your response and for sharing your concerns. I completely understood those points when I initially wrote to you, which is why we remain committed to collaborating with the TinkerPop team to finalize the match() step specification.
Our reasoning is that any specification should be grounded in solid implementation and testing. By prioritizing the implementation, we can uncover potential pitfalls that may not be apparent from a purely usability-oriented perspective. Additionally, I would like to note that the query specification cannot be finalized without the context of the transaction processing lifecycle. Discussions in that area have stalled, and I would appreciate it if we could move forward at a faster pace. We have significant practical experience in this area and are eager to contribute. Best regards, Andrii Lomakin On Fri, Jan 9, 2026 at 11:37 PM Cole Greer <[email protected]> wrote: > Hi Andrii, > > Thanks for providing your update. I understand your decision, although I > regret that the community may not be as involved in the development > process. My main concern is that if you and your team complete all of the > GQL work independently, that the final product might not be as aligned with > the whole TinkerPop community, which may complicate the process of > officially incorporating the changes into Gremlin. > > I understand your need to develop this feature at your own pace and > without added constraints. I would like to ask if you and your team can > continue to keep us informed as you proceed. I'd be happy to keep providing > early feedback on any decisions and progress made by your team, in hopes > that we can maintain alignment during development. > > Thanks, > Cole > > On 2026/01/09 08:10:06 Andrii Lomakin via dev wrote: > > Dear TinkerPop team, > > > > I would like to provide an update on our next steps regarding the > > integration of GQL with TinkerPop. > > > > We have hired a new developer, Sandra Adamiec, who will focus on > > implementing GQL in YTDB (in CC). After several internal discussions, we > > have concluded that our initial approach—creating a detailed > specification > > before tailoring the implementation—carries significant DevX and > > performance risks. > > > > Instead, we have decided on the following path: > > > > 1. Create the GQL implementation and its Gremlin integration within our > own > > distribution. > > 2. Once functional, collaborate with the TinkerPop team to finalize a > > match() step specification that works for both parties. > > 3. Port the implementation (or relevant parts of it) to the official > > TinkerPop distribution. > > > > This strategy allows us to decouple our product timeline from the > TinkerPop > > release schedule. Additionally, it helps mitigate the performance and > > functional risks we identified in the roadmap for the new Gremlin server > > implementation. > > > > We look forward to continuing our discussions regarding the match() step > > specification once we have a functional foundation to use for validation. > > > > Best regards, > > Andrii Lomakin > > YouTrackDB development lead > > > -- Andrii Lomakin YouTrackDB development lead
