CLIENT-4878 serverside ael parsing - #60
Conversation
…g. requires fixing
paul-brackin
left a comment
paul-brackin
left a comment
There was a problem hiding this comment.
Looks like there could be some issues, but first please rebase, resolve conflicts, and get us all green on the tests.
Re-ran the failing test and it went fine. I had merged |
|
I am okay with that. I had split it to 2 PRs because of larger changes and the #71 PR is still dependent on PAC (aerospike/aerospike-client-python-async#85) and rust (aerospike/aerospike-client-rust#221) changes and cannot be merged before those. I have changed the base of #71 to |
|
Thanks for retargeting. I'm closing #60 as absorbed; review moves to #71's combined diff. And #71 can merge ahead of those 2 PAC & core PRs. The dependency is in the pin, not the code. Your own guards already make every #85-dependent surface unreachable on today's PAC. You'd just need these 2 additional changes in #71:
|
|
Closing as absorbed into #71 |
… selection (dark-launched) Field-43 server-compiled AEL and field-44 server-side index selection, merged dark: both feature gates in aerospike_sdk/feature_gates.py are hard-false, so dev behavior is unchanged. Flip requires: PAC #85 (honest version gates) + core #221 merged, a released server with the query optimizer, an 8.1.3+ CI leg, and the client/server AEL equivalence corpus green. Absorbs PR #60.
2 participants
The SDK layer changes that stays on top of aerospike/aerospike-client-rust#203 and aerospike/aerospike-client-python-async#77
Tests have been updated to do both client side and server side AEL parsing.
Many of the server-side AEL parsing fails - a work in progress on server side.