rossmohax

It does, but AFAIk only for prepapred queries and cache is local to the backend serving query. See `plan_cache_mode` param.

Please use neither. Reason is as number of lookups varies, you'll have to template query as a string adding either values to IN list or `OR` conditions. Not only it causes excessive string concatenation on your app side, these queries are also seen as distinct queries which has following drawbacks:

- every query has to be planned

- your DB driver can't prepare query as they are all different

- collecting per query stats becomes nightmare if number of arguments per query varies in wide range. metrics cardinality is a problem.

Correct way to handle it is pass all args as single parameter of type array and use `= ANY($1)` or if there are multiple columns build a virtual table and join:

  SELECT a,b FROM table 
  NATURAL JOIN ROWS FROM (
    unnest($1::type_of_a[]),
    unnest($2::type_of_b[])
  ) t(a,b)

There was a company I can't recall its name, but they took VC funding for one round, retained voting control, built successfull business and then refused to go for another VC round. VC still owns company, but can't exit and founders are happy with the way things are.

Does ZFS or any other FS offer special operations which DB engine like RocksDB, SQLite or PostgreSQL could benefit from if they decided to target that FS specifically?

Life would be easier if UK switched driving sides.

this will work 0% of the time if app closes config after reading

This talk is must see to understand what SI tries to achieve.

I like the appeal of model being bidirectional. Also modelling sequence of actions is really not solved problem in Terraform & Pulumi: canary change, check metrics, rollout to the rest of the region, check metrics, then all regions, if they solved it all while being "declarative" and high level it can be the next tool of choice for me.

I am not worried about UI representation of the model like many comments, it is not the main point of this project as I understand. UI just that - a representation, same relationships might as well be coded in HCL or the like of it.

This project could borrow a lot from unikernels. If they mandate running it as a VM, there is no HW to support.

TLDR; it is a Google search proxy for Mullvad customers with 50 searches per 24 limit. There is an option to do unlimited number of searches in their cache.

TLDR; hash the hash with fast function, like fibonacci hash which is just single multiplication by a constant, then take lower bits as your bucket index.

How does Mongolian IDE look like?

There is still no good support for io_uring in Rust. AFAIK none of existing solutions fully support multishot ops, zerocopy ops with delayed buffer release, sending messages to another ring with IORING_OP_MSG_RING or convenient ops linking

I thought Witcher 3 had a good story until I played Disco Elysium.