BLOCK_SIZE C Macro collision

[pedroerp]

What happens?

Not a bug in duckdb, but we recently bumped into a collision issue in one internal system which ends up including a kernel header (fs.h) that defines BLOCK_SIZE as a C macro, and that ends up conflicting and rewriting Storage::BLOCK_SIZE in duckdb, and thus failing compilation.

I know there are hacky ways to around this (like undefining this macro before including duckdb), but considering BLOCK_SIZE is a fairly general name, any chance you would be willing to accept a PR that renames Storage::BLOCK_SIZE to something like Storage::DUCKDB_BLOCK_SIZE or maybe Storage::block_size?

To Reproduce

Include any library that has a C-style define BLOCK_SIZE before including duckdb.hpp.

OS:

Ubuntu, but reproduces on any.

DuckDB Version:

Any

DuckDB Client:

Any

Full Name:

Pedro Pedreira

Affiliation:

Meta Platforms Inc.

Have you tried this on the latest master branch?

• I agree

Have you tried the steps to reproduce? Do they include all relevant data and configuration? Does the issue you report still appear there?

• I agree

[carlopi]

Hi @pedroerp! I took a look at the issue, and I think I have an alternative solution that might require less changes: move declaration of struct Storage to an 'inner' header, that should not be included from duckdb.hpp.

Could you potentially check if this is of some use, doing something like:

cd duckdb
git remote add carlopi https://github.com/carlopi/duckdb.git
git fetch carlopi
git checkout carlopi/includeStorage

?

[ Here the linked PR: #6505 ]

[pedroerp]

Hi @carlopi thanks for the quick reply. Looks like this solved the issue. Any chance you could get it merged?

[github-actions]

This issue is stale because it has been open 90 days with no activity. Remove stale label or comment or this will be closed in 30 days.

[github-actions]

This issue was closed because it has been stale for 30 days with no activity.