@theburningmonk I wonder why AWS can skip the eventual consistency check when creating resources but MUST be consistent to when reporting a CREATE_COMPLETE.
Perhaps it's so that AWS can be sure that the CLI/SDKs will always see the created resources after the stack reports CREATE_COMPLETE.
This new CFN feature sounds incredible! Hopefully, it can now compete with Terraform. https://t.co/8ZzNIppcOW
I suspect this is the feature that caused this bug. 🐜
It's wild that there has been an S3 / CFN bug causing stack creation failures around 5% of the time for > 60 days. It blows my mind how this was not picked up during some kind of internal testing. ESPECIALLY given how big both of these services are. https://t.co/P6fWam7uS6
Ever wondered how isolation levels work? How transactions are implemented?
I’ve spent the last few months going deep and @andy_pavlo’s CMU course is very approachable and is the best of the best. Also what’s even better is it’s free. https://t.co/y4ytVSfAQO
@petehanssens Rust solves a similar set of problems as C++, operating systems, drivers, UI, games etc. Excluding some games companies Australia does not write any software that needs this level of optimisation.
It's wild that there has been an S3 / CFN bug causing stack creation failures around 5% of the time for > 60 days. It blows my mind how this was not picked up during some kind of internal testing. ESPECIALLY given how big both of these services are. https://t.co/P6fWam7uS6
I was thinking that if you submit an entire transaction in 'one go' it would be treated as a statement. Though as far as pg/pgbouncer is concerned it doesn't matter if it's in 'one go', if it's go a BEGIN; pgboncer will drop the whole connection.
Hey @jpcamara -- enjoyed your article. https://t.co/wfjJQEpnU1.
Question -- if you submit a transaction as a single query, e.g. the `BEGIN; statements; COMMIT`; in one go, can you still use statement pooling?
@bulkanevcimen@petehanssens 3/ I know you didn’t ask, but I’ll share anyway. Where I’m starting is building a redis cone, one thread for the database, a thread per connection, and then MPSC channels for sending the request to the DB and a one shot channel for sending the response back.
@bulkanevcimen@petehanssens 2/ If you really want to flex the language building very simple MVCC database a great learning experience. You get, file serialisation, networking, concurrency primaries, memory synchronisation, threading, networking, some mutability.
@bulkanevcimen@petehanssens I was thinking of the same, though, ended up getting carried away on side projects. I love me some result / optional type, so good.