Bug: Segfault when dimensions of added vector don't add up (Rust) about usearch HOT 8 OPEN

I will look into it tomorrow, maybe this error happens with other functions too...

from usearch.

@MarkReedZ, check #413 🤗

from usearch.

Nice! Any chance you could open a PR?

from usearch.

Another reason for segfaults is trying to add to the index when its capacity is 0 (not reserving). This could also be checked in the add function.

from usearch.

This one is harder to justify as it's not a zero-cost change. The size is an atomic variable, and checking it from many threads would cause contention.

from usearch.

I'm taking a look at this. Currently the rust tests are failing on add.

test tests::test_add_remove_vector ... thread 'tests::test_add_remove_vector' panicked at rust/lib.rs:1434:9:
assertion failed: index.add(id2, &second).is_ok()

I added a test for the seg fault example however this passes for me.

index.add(row as u64, &[3, 5]); //2 is not 192 and causes a segfault

I've added tests for other permutations of add and have hit a hang.

from usearch.

I'm looking at an unrelated error and cleaning up the tests to be more rust idiomatic.

However, looking at the PR - are you suggesting fixing this in the base library by adding a size_t length argument to index::add then checking to ensure it matches options.dimensions? Or only fixing this for rust by updating rust/lib.cpp?

from usearch.

I meant changing only the rust/lib.cpp 🤗

from usearch.