SQL Indexing

[Abstrct]

Is the ability to index directly into a postgres database available in Gordian or is this a component that is different compared to comtBFT?

[mark-rushakoff]

Hi @Abstrct, thanks for opening our first community question.

Are you asking about indexing Cosmos SDK transactions? gcosmos does not provide that yet. It looks like we are going to need to provide transaction indexing very soon, for basic transaction querying and for IBC support -- so it's very high on our priority list.

This will end up looking similar to the tmstore interfaces in Gordian. We will get the interfaces in place first with an in-memory implementation for testing, and once that is developed enough, we will provide an implementation in SQLite.

We don't have immediate plans to provide any Postgres or MySQL implementations of storage layers, but we anticipate that at least one community-provided implementation will pop up sooner or later. At that point we could consider accepting their implementation under the gordian-engine organization.

If no community implementation happens, we would probably eventually write one ourselves, but that is very low priority as of today.

[Abstrct]

Thanks! I was specifically asking about the Cosmos SDK transaction indexer, but I appreciate about the added information about the indexing of storage layers down the road too.

We rely on the postgres transaction indexer pretty heavily in Structs, so that would be a blocker for our small project in terms of testing Gordian. That said, I've always had fairly strong opinions on possible ways the CometBFT postgres implementation could have been improved. Is this forum the best place to participate in conversation on the topic going forward?

[mark-rushakoff]

Is this forum the best place to participate in conversation on the topic going forward?

Yes, we are intending to use this GitHub Discussions board as the primary location for discussing general development of Gordian. The board is open for anyone to read, and chances are anyone who wants to post already has a GitHub account. Moreover, we expect that the search here will work well enough, and I assume search engines will index the discussions.

If you have a specific feature request, an issue in the appropriate repository is a better fit, as issues are intended to be more focused on specific work items. But it sounds like you have more of a general vision in mind. Feel free to continue this thread or start a new one.

I did spend some time researching and planning what an indexer would look like on our side, and it doesn't seem too bad. It also looks like there are more projects relying on the Comet Postgres indexer than I expected, so API compatibility at the Postgres table level is probably more valuable than I initially expected too, so we could probably justify working on that sooner than I suggested in my previous post. Early feedback and suggestions are welcome.