Jason Thorpe <thorpej%me.com@localhost> writes: >> On May 21, 2018, at 8:38 AM, maya%netbsd.org@localhost wrote: >> >> I do agree that having those functions in libc was a bad idea, > > So, if we think having them in libc is a bad idea (I don’t disagree, > but the MD5 stuff at least has been there for a long time), and if we > prefer people using the OpenSSL variants, then it seems like a > palatable solution is to disable installing the man page for the libc > variant (or perhaps rename it to have a libc_ prefix). Heck, we can > even rename the libc variant functions to libc_whatever and export > alias symbols for the legacy names to keep binary compatibility. Thinking about minimal fixes, it seems no longer installing Mail is not a big deal; using it borders on retrocomputing. I only type Mail out of habits dating to probably 7th Edition, not realizing that mail has caught up until this thread. As for md5/MD5, it seems easy enough to install the libc md5 man page as libc_md5 and to add a SEE ALSO in the openssl version. Arguably both pages would have an pointer to the other becuase this is actually confusing, not just a fs case issue. I am guessing that on -current, the openssl man pages have arrived - I don't find them on my -8 system. On a RPI3 running slightly old -8, I see md5(3), which describes functions starting with MD5. So it seems that th md5(3) man page could just be removed, with openssl's MD5(3) page having a crossreference to NetBSD's MD5Init.
Attachment:
signature.asc
Description: PGP signature