Tom White

Thank you for the suggestions @martindurant! I tried using cramjam's `Buffer` to wrap a NumPy array, but it seems to copy the bytes on construction. This also happens when using...

Thank you for working on this @milesgranger! I used `copy=False` and it now avoids a copy (verified using memray). But using it in s3fs still has problems, and unfortunately the...

> [@tomwhite](https://github.com/tomwhite) : it makes it look like a network correction fault; just checking that writing a `bytes` does still work? Yes, that still works. ``` >>> fs.pipe("s3://cubed-unittest/mem-array/s3test", b"hello") #...

Not sure if this is related: ``` >>> import cramjam >>> import numpy as np >>> data = cramjam.Buffer(np.frombuffer(bytearray(b"hello"), dtype="uint8"), copy=False) >>> len(data) 5 >>> data.read(10) b'hello\x00\x00\x00\x00\x00' ``` Shouldn't it...

> Figured it out, I have `fsspec.filesystem('s3').pipe(...)` w/ a zero-copy cramjam Buffer working, can install version `2.11.0rc2` Can confirm that it works now. Thanks @milesgranger! ``` >>> import cramjam >>>...

Thanks for looking at this @HaoZeke! I'm not working on this, so I'm happy for you to continue. I think while investigating it's OK to [force the GIL to be...