Open BugenZhao opened 2 years ago
@BugenZhao Hi,I want try this. May i ask, should we update src/sqlparser
which mentioned in #2602 firstly? If sure, how to update.Directly copy paste? And is the repo andygrove / sqlparser-rs
or sqlparser-rs / sqlparser-rs
?
@BugenZhao Hi,I want try this. May i ask, should we update
src/sqlparser
which mentioned in #2602 firstly? If sure, how to update.Directly copy paste? And is the repoandygrove / sqlparser-rs
orsqlparser-rs / sqlparser-rs
?
Cool! cc @neverchanje Do we have a plan for syncing sqlparser from the upstream? Besides, this may not just be simply parser stuff. We may need to do some work on the optimizer as well. :)
Cool! cc @neverchanje Do we have a plan for syncing sqlparser from the upstream? [...]
I think having a plan how to incorporate changes from sqlparser-rs into our project would make sense. I did another bug recently (see https://github.com/risingwavelabs/risingwave/pull/5673), which turned out to be already fixed in upstream. Do you think it makes sense to use the parser as a git submodule?
There was a discussion on slack and the conclusion is to diverge from upstream: https://risingwave-labs.slack.com/archives/C034U0NH5ND/p1661152590242349?thread_ts=1661152070.390169&cid=C034U0NH5ND
I personally suggest that we can diverge from their branch and evolve it based on our own needs. sqlparser-rs is not really a community-oriented project. It supports too many dialects while none of them is well supported. Thus none of the database communities can benefit from this project other than forking a stand-alone branch. That’s what MZ is doing. MZ’s parser was originally forked from sqlparser-rs but now its code is way different from the upstream.
Unnesting.
Subquery on the right side of multi-assignment.
UPDATE ... SET ... FROM
.Related PR: #2602