sergej-klimenko

I have access to both Intel and ARM Macs, but need some time to upgrade them to 15.0. I hope I can check it out in the next few days.

Tested it on ARM64 M1 Pro 15.0.1 with [instantclient-basic-macos.arm64-23.3.0.23.09-1](https://download.oracle.com/otn_software/mac/instantclient/233023/instantclient-basic-macos.arm64-23.3.0.23.09-1.dmg) Run Oracle DB Free ``` docker login container-registry.oracle.com docker pull container-registry.oracle.com/database/free:latest docker run --name oracle \ -p 1521:1521 \ -e ORACLE_PWD=Test123...

But found something strange. Stop listener in running container ``` docker exec -it oracle lsnrctl stop LSNRCTL for Linux: Version 23.0.0.0.0 - Production on 11-OCT-2024 08:28:04 Copyright (c) 1991, 2024,...

I discovered something strange in the debug output of test. None of the lines like ``` handlers.go:172: 20:33:59.357 DBG "Object.Close" attrs={"object":"0x600001b33570"} ``` are followed by "Object.Close close sub-object". However, they...

> Try with using the same s (pslice) for each call - that'll eliminate any Go-side allocations. Unfortunately, that changes nothing - just need slightly more time to double the...

> Ok, without logging in SetAttribute, it's way better: I've got more or less same results about RSS as before this commit... ``` === RUN TestPlSqlNestedObj z_plsql_types_i390_test.go:114: Alloc: 0.485 MiB,...

Also tried to pass parameter as IN-only with no difference...

Maybe you can try to run this test with AddressSanitizer (unfortunately not supported on my arm64) ? GO_LDFLAGS='-fsanitize=address' CGO_CFLAGS='-O0 -g3 -fsanitize=address' go test -asan -run "TestPlSqlNestedObj" -v

I've translated this test one-to-one to python-oracledb (using thick mode) - there is no memory leak here.

Perhaps you meant the memory grow stays around 10% with justOut , not with justIn ? Because in my run : justIn - 380%, justIn < 10% ``` z_plsql_types_i390_test.go:152: TestPlSqlNestedObj/justIn:...