Repository navigation
[GLib][Ruby] Make underlying calls to FileSystemFromUri() on gs/gcs filesystems work #55284
Description
Describe the enhancement requested
On the following PR we moved GCS to its own shared library:
Ruby bindings are currently not linking libarrow_gcs.so nor registering the filesystem with LoadFileSystemFactories. This issue is to track making underlying calls to FileSystemFromUri() on gs/gcs filesystems work.
This spawns from these comments:
Component(s)
GLib, Ruby
Activity
[-][CGLib][Ruby] Make underlying calls to `FileSystemFromUri()` on `gs`/`gcs` filesystems work[/-][+][GLib][CGLib][Ruby] Make underlying calls to `FileSystemFromUri()` on `gs`/`gcs` filesystems work[/+][-][GLib][CGLib][Ruby] Make underlying calls to `FileSystemFromUri()` on `gs`/`gcs` filesystems work[/-][+][GLib][Ruby] Make underlying calls to `FileSystemFromUri()` on `gs`/`gcs` filesystems work[/+]Code-ve commentedon Oct 29, 2026
Hello @raulcd, I'd be happy to work on this problem.
Based on the issue description and the changes made in #54905, my impression is that moving GCS functionality into the standalone
libarrow_gcs.solibrary might have created a situation where the Ruby bindings lack the proper linking or filesystem factories registration. This could lead to failed resolution of GCS filesystems forFileSystemFromUri()withgs://orgcs://URIs.It would be good for me to find out how the Ruby bindings link to Arrow and Arrow GLib libraries, what loading mechanism of filesystem factories is used, and what changes are needed to allow the Ruby bindings support GCS the same way as in the new library configuration. I would also need to add some regression tests for both schemes in order to make sure about the proper behavior while making these tests independent from real cloud credentials as much as possible.
Would it be alright for me to tackle this issue? Are there some implementation or testing specifics I should consider before getting started?
raulcd commentedon Oct 29, 2026
Hi @Code-ve thanks for the interest on this issue. As this would be your first contribution to the project and this one requires quite a lot of knowledge about our Ruby binding internals I'd rather prefer for a current Ruby bindings maintainers to follow up instead of heavily relying on AI tools in order to work on it. Thanks.
pinging @kou as the main Ruby bindings maintainer as I think he wanted to work on this himself.