Build fails on GCC 8 due to missing -lstdc++fs when using std::filesystem

[liuchuang001]

What happens?

When building pg_duckdb on systems using GCC 8 (e.g. CentOS 8), the PostgreSQL extension fails to load due to unresolved std::filesystem symbols.

GCC 8 supports std::filesystem but requires explicitly linking against libstdc++fs. This is no longer necessary in GCC 9+, where filesystem was merged into the standard C++ library.

To Reproduce

On a system with GCC 8.x (e.g., CentOS 8):

Build pg_duckdb using the current Makefile

Load the extension in PostgreSQL:

sql

LOAD 'pg_duckdb';

You will encounter an error like:

undefined symbol: _ZNSt10filesystem7__cxx114path14_M_split_cmptsEv

Diagnosis:

Inspecting pg_duckdb.so with nm -D shows unresolved symbols related to std::filesystem:

nm -D pg_duckdb.so | grep filesystem

# Output:
# U _ZNSt10filesystem7__cxx114path14_M_split_cmptsEv
# U _ZNSt10filesystem18create_directoriesERKNS_7__cxx114pathE

These symbols are undefined unless -lstdc++fs is explicitly passed during linking on GCC 8.

Suggested Fix:

Modify the Makefile to detect if GCC version is 8 and add -lstdc++fs accordingly:

GCC_VERSION_MAJOR := $(shell g++ -dumpversion | cut -f1 -d.)
ifeq ($(shell [ $(GCC_VERSION_MAJOR) -eq 8 ] && echo true),true)
  PG_DUCKDB_LINK_FLAGS += -lstdc++fs
endif

This keeps the build compatible across all compiler versions:

GCC < 8: no change (but std::filesystem isn't usable anyway)

GCC 8: adds required -lstdc++fs

GCC ≥ 9: safely ignored

Let me know what you think! Thanks for all the amazing work on pg_duckdb! 🙌

OS:

linux centos8 x86

pg_duckdb Version (if built from source use commit hash):

12dee2e43520144059f007d37ce4863b9c879d31

Postgres Version (if built from source use commit hash):

16.9

Hardware:

No response

Full Name:

liu chuang

Affiliation:

[email protected]

What is the latest build you tested with? If possible, we recommend testing with the latest nightly build.

I have tested with a source build

Did you include all relevant data sets for reproducing the issue?

Yes

Did you include all code required to reproduce the issue?

• [x] Yes, I have

Did you include all relevant configuration (e.g., CPU architecture, Linux distribution) to reproduce the issue?

• [x] Yes, I have

[JelteF]

Sounds like a reasonable fix. If you create a PR I will merge it.