Nick Moore

G'day @bksp for what it's worth I *still* think this is a great hardware feature to have exposed in MicroPython :-) The work in this repo has now been merged...

Thanks for letting us know. I'll go back and build some older versions and see where they diverged. But yes, I can confirm that this is a problem, I get...

f0561abc changes CONFIG_MBEDTLS_HARDWARE_SHA in the process of updating to a newer esp-idf. Reverting it and rewinding to espressif/esp-idf@4ec2abb fixes the problem.

Hmmm, thanks for the info @projectgus, it sounds like a quick fix anyway. @dpgeorge, if you're busy with other things I'll push a PR tonight ... PS: If the hardware...

OK, so microseconds after I pushed a fix for this I accidentally did this and discovered something a little weird: ``` >>> import binascii, hashlib >>> hash = hashlib.sha1(b'mango') >>>...

(aha, it's doing the padding step as part of the _finish call, so that's getting done twice and scrambling the results. The only way around it would be to keep...

Yeah, I'd rather it worked like the CPython version and just return the same answer every time, but throwing an exception would be a lesser evil than returning wrong results....

Yeah, I'd like to push nicer behaviour upstream to main micropython. Perhaps just by calling `mbed_tls_sha1_clone` before `mbed_tls_sha1_finish`. That would allow multiple calls to digest and partial digests and bring...

> I don't know of the exact policy, but after this bug should not a > sanity check on the> answer be included in the tests here? Yes.  I'll put...

I turned some debugging on ... lwip_send is getting error -5 back from netconn_write_partly and then translating that to EINPROGRESS. The trouble is that EINPROGRESS isn't consistently defined: ``` esp-idf/components$...