(I still need to fixup python_hacl_namespaces.h but first want to make sure this PR works.)

@gpshead these are just build errors, apparently I didn't export the right CFLAGS somewhere

before I dig further into this, would you like all of the hacl files to be bundled into a single libhacl.a? or is it ok to rely on the existing build system to build single .o files for the non-sha2 algorithms?

before I dig further into this, would you like all of the hacl files to be bundled into a single libhacl.a? or is it ok to rely on the existing build system to build single .o files for the non-sha2 algorithms?

I think what you've done in ths PR should work fine. one .c file for each of md5 and sha1, no specific need for a library.

from a build simplicity perspective with the way we have things set up in our plethora of build configs, the lesson from the sha2 stuff was not to reuse the same .c/.o/.a with non-static symbols in multiple extension modules. (i may not have needed the .a for sha2 in the end but i decided to leave it in place when merging sha256 and sha512 into sha2 - if we don't need it setup that way in the end it could be undone later to just be the .c file as these are)

Ok looks like we're good now. I'll get started on SHA3 while I wait for this one to land.

🤖 New build scheduled with the buildbot fleet by @gpshead for commit b3dd4fd 🤖

If you want to schedule another build, you need to add the 🔨 test-with-buildbots label again.

🤖 New build scheduled with the buildbot fleet by @gpshead for commit b3dd4fd 🤖

If you want to schedule another build, you need to add the 🔨 test-with-refleak-buildbots label again.

I'm not sure what to make of the buildbot run. Three failures are in test_zipfile, another one in test_asyncio, all seem unrelated to this PR.

I'm not sure what to make of the buildbot run. Three failures are in test_zipfile, another one in test_asyncio, all seem unrelated to this PR.

nothing to worry about. those look like flaky / poorly designed tests and/or infrastructure issues.