Essentially we don't want to - it's a major undertaking, and showing how Zarr compares with file-based approaches isn't the point of the paper. We show that Zarr is competitive in various dimensions using the simulations, and show that it is practical on real data in GEL.
Also we're definitely not going to run Genozip because of its egregious licensing conditions.
We do need some short and clear way of saying this for the paper, just to preempt the obvious reviewer request of "hey, it'd be interested to see how X does on GEL!".
Essentially we don't want to - it's a major undertaking, and showing how Zarr compares with file-based approaches isn't the point of the paper. We show that Zarr is competitive in various dimensions using the simulations, and show that it is practical on real data in GEL.
Also we're definitely not going to run Genozip because of its egregious licensing conditions.
We do need some short and clear way of saying this for the paper, just to preempt the obvious reviewer request of "hey, it'd be interested to see how X does on GEL!".