Feature request: `concatenate` for `NDArray` (C-API/Python) without decompression

[rivershah]

Request native concatenate functionality for blosc2.NDArray, exposed in both the C-API and the Python wrapper.

Requirements:

• C-API: Provide a C function to concatenate compressed b2nd along a specified axis.
• Python API: Wrap the C function, mimicking the numpy.concatenate signature (sequence of arrays, axis parameter). Ref: NumPy Docs
• Core Constraint: Must operate directly on compressed data, avoiding full decompression/recompression.
• Assumptions: Input arrays share identical compression settings (codec, clevel, cparams, etc.).

Use Case:

Efficiently join large, pre-compressed datasets in both low-level C applications and Python without the performance penalty of decompression/recompression. Exposing it in the C-API is key for broader integration.

blosc2.SChunk.insert_chunk is very fast. Can this feature use this, along with metadata updates?

Thanks for considering this.

[FrancescAlted]

Thank you for your feature request! Due to its complexity, we recommend either contributing via a PR or supporting the project through a donation/contract to prioritize its development. Please see the Support Blosc for a Sustainable Future section for more details.

[rivershah]

Thank you for your great work. Fully agreed with this model and I've done so.

[FrancescAlted]

Thank you! We have received the donation and should start work on this as soon as possible (typically not before 2 weeks, but no later than 4 weeks). Stay tuned!

[FrancescAlted]

Ok, our agenda looks better now. Next week we plan to have a try at this one. Thanks for your patience.