Thanks for your useful comments Karel! (inline) On Wed, Dec 7, 2016 at 2:03 PM, Karel Zak <kzak@xxxxxxxxxx> wrote: > On Wed, Dec 07, 2016 at 10:40:36AM -0500, Whilom Chime wrote: >> I've worked up a simple patch to libuuid to allow for generation of >> MD5/SHA-1 UUIDs (DEC V3, V5 conformant) since I can't seem to find any >> other C-level APIs to that effect and libuuid seems to be the C-level >> "go-to" for such. Does that seem reasonable? > > What about http://www.ossp.org/pkg/lib/uuid? I see this library in > Fedora and it supports all the UUID variants. The util-linux > (originally e2fsprogs) libuuid is about time based UUID only. > > I'm have real doubts we need to support another UUID variants in > our libuuid. > Haha! Wow, libossp-uuid.a would have satisfied me if I had of found it... A bit like a sledgehammer to tap a tack, but I agree, given that there *is* a C API out there it's prolly best to leave util-linux'd libuuid alone. Though I really wish md4 and sha-1 UUID's had gotten in there in the first place, it's literally forty lines of code :P I will either use libossp-uuid.a or just slip the SHA-1 version into my code myself, I won't pursue a patch to util-linux. But one more process question below, just so I understand for future... >> Obviously this requires openssl, not a dependency that currently seems >> to exist in util-linux. Is adding that depency reasonable? I've >> added it as a default-yes ./configure --with-openssl option. > > Hmmm... for the hash? We have MD5 in the tree (it would be probably > possible to have there SHA-1 too), although for some certifications > (FIPS) and consolidation is probably better to use external > verified library. > >> "Stupid" question: to expose the two new APIs it looks like I need to >> add something like: >> >> UUID_2.29 { >> global: >> uuid_generate_md5; >> uuid_generate_sha1; >> } UUID_2.20; >> >> to libuuid.sym but I'm completely unclear as to what version I should >> use, the current 2.29, some imagined 2.30, 2.29_rc3...? > > We do not use _rcN there, it's X.Y release number, during -rc time > it's unstable and modifiable. > Could you just help me understand how lib??.sym works? If I have 2.29 source and I add UUID_2.30 {...}... does that mean that the symbol would be added in my local 2.29 build for testing, but once my hypothetical patch was accepted the "world" would not see my new APIs in any version less than 2.30? Or does the package maintainer have to manually maintain all kinds of lib??.sym files as revisions happen... Thanks! >> And there may be other things I need to do to add APIs? > > Maybe docs. > >> CRYPTO_LIBS= >> AC_ARG_WITH([openssl], >> AS_HELP_STRING([--with-openssl], [compile with OpenSSL support]), >> [], [with_openssl=yes] >> ) >> if test $with_openssl != no >> then >> a=0 >> AC_CHECK_HEADERS(openssl/md5.h, [], [a=1]) >> AC_SEARCH_LIBS([MD5_Init], [crypto], >> [if test x"$ac_cv_search_MD5_Init" != x"none required"; then >> CRYPTO_LIBS="-lcrypto"; >> fi], [a=1]) >> >> if test $a == 0 >> then >> AC_DEFINE(HAVE_MD5, [1], [Define to 1 if you have MD5 functions]) >> fi >> >> a=0 >> AC_CHECK_HEADERS(openssl/sha.h, [], [a=1]) >> AC_SEARCH_LIBS([SHA1_Init], [crypto], >> [if test x"$ac_cv_search_MD5_Init" != x"none required"; then >> CRYPTO_LIBS="-lcrypto"; >> fi], [a=1]) >> >> if test $a == 0 >> then >> AC_DEFINE(HAVE_SHA1, [1], [Define to 1 if you have SHA1 functions]) >> fi >> fi >> AC_SUBST([CRYPTO_LIBS]) > > The best is to use pkg-config, see how we use PKG_CHECK_MODULES in the > configure. > > Karel > > -- > Karel Zak <kzak@xxxxxxxxxx> > http://karelzak.blogspot.com -- To unsubscribe from this list: send the line "unsubscribe util-linux" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
