Xing Guo

[WIP] Improve gp_hyperloglog codes and add tests for it.

1

This patch helps remove redundant base64 encoding and decoding functions in gp_hyperloglog.c and add test cases for gp_hyperloglog. 1. Postgres upstream provides the implementation of base64 encoding and decoding utilities....

Add cargo-readelf subcommand.

1

Hi, I found that the `cargo-readobj` command proxies the `llvm-readobj` tool and uses `--elf-output-style=GNU` to make the output compatible with `gnu-readelf`. LLVM do provide a tool called `llvm-readelf` which is...

Fix incorrect hash-key of Redistribute-Motion when creating path for multi-DQA expr.

4

This patch is to fix #14096. How does multi-DQA work: When creating path for multi-DQA (multiple distinct qualified aggregate), the planner generates the plan as follows: ``` Finalize Aggregate ->...

Collapse the pythonsrc-ext submodule.

2

We're going to build gpdb on runway/buildweb in the future, on which we are unable to fetch submodules from outside repos. This patch helps collapse pythonsrc-ext submodules. This patch also...

GPG keys error

1

Importing GPG key 0x8C671CDF: Userid : "Zhang Sen (Builder of Senorsen) " Fingerprint: 67AD B50C 6FE6 F3D5 1AD7 078E 71BB CDA2 8C67 1CDF From : /etc/pki/rpm-gpg/RPM-GPG-KEY-SENORSEN Is this ok [y/N]:...

Add new option --gp_version to pg_config.

2

Some extension development framework need to read pg_config at compile time to determine the Greenplum version. This patch adds a new option --gp_version to pg_config to allow these tools to...

Managing GUC parameters in different places is hard to maintain. This patch organizes GUC definitions in a single place. Also, we use define_xxx_guc() APIs to define these parameters and it...

This patch introduces a new GUC parameter pgml.omp_num_threads to control control the number of threads that OpenMP can spawn. This PR contains 2 commits. The first one is refactoring the...

PostgresML is spawning too many threads.

11

Recently, we found that PostgresML was spawning too many threads and threads is waiting for each other. No deadlock but low performance. ## Root cause The root cause is that...

Cannot use `pgml.activate_venv()` to set environment for parallel workers.

8

Currently, `pgml` provides a UDF called `pgml.activate_venv()`[^1]. However, when a query requires parallel workers, the venv environment cannot be set for parallel workers. This is not very easy to reproduce...