-
Notifications
You must be signed in to change notification settings - Fork 2.7k
truly optional rocksdb #14117
truly optional rocksdb #14117
Conversation
|
bot rebase |
|
Rebased |
niklasad1
left a comment
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
nice, looks good
Co-authored-by: Niklas Adolfsson <[email protected]>
|
Are the Could I get another review :) |
|
bot rebase |
|
Rebased |
It's probably just because there were a bunch PRs merged recently that broke that, rebased the branch and it's seems to be okay but cumulus is currently broken because paritytech/cumulus#2574 So nothing to worry about |
And why are you using |
I am aware, which was my first quarrel with it. Now we had the gcc/clang update as part of our build container and again rocksdb made em all red. On the other handside, not all prebuilts can be linked with rocksdb due to missing symbols and friends. Time not worth spending, hence the radical option of cutting it out entirely.
We are not. |
What does it do?
Currently the
sc-client-dbis pulled in with default features at the levelcliwhich makes it a defacto dependency for all parachains. With the latest slew of compiler releases not include<cstdint>by default anymore, androcksdbbinding a significant amount of CI resources, I felt it was time to get rid of it without having to continually patch substrate.What important points should reviewers know?
RocksDB is now optional two levels further up, behind another
rocksdbfeature flag defaulting to on. The default behavior should not have changed.Is there something left for follow-up PRs?
node-testingmakes some assumptions of having some identifiers around which are not upheld, it surfaced when testingno-default-features, but otherwise no change.We are successfully using this with
polkadot-v0.9.41(plus a few patches) and clang-16 and gcc-13.