On Tue, Feb 11 2020, Paco Esteban <p...@e1e0.net> wrote: > On Sat, 25 Jan 2020, Paco Esteban wrote: > >> Hi ports@, >> >> Here's a new port for git-crypt, which is a tool for transparently >> encrypt files on git repositories (so one can have sensitive information >> on remote repositories). You can find more info here: >> >> https://www.agwa.name/projects/git-crypt/ >> >> I decided not to include gnupg as a dependency. The software can use >> pgp keys (for asymetric) or openssl (for symetric) encryption. The >> build clearly depends on openssl (it seems to work fine with libressl as >> far as I can see), but for pgp it calls the gpg binary, so it does not >> depend on it for building. Please correct me if this is wrong.
This makes sense. Note that security/gnupg provides a gpg executable, security/gnupg2 does not. Let's hope that git-crypt doesn't require a more recent gpg version. >> Another thing is that the man page generation leaves the resulting >> man page file on WRKSRC/git-crypt.1 but it should leave it on >> WRKSRC/man/man1/git-crypt.1 and I do not know why (when doing things >> manually outside of the ports infrastructure it works as expected), >> that's why I added the pre-fake hack. If anyone knows a better way to >> deal with this, please tell me. Ah ha. No idea what's happening. > ping ? > > (port attached for convenience) New tarball attached: - merge V and DISTNAME - zap CONFIGURE_STYLE - move WANTLIB above MASTER_SITES (Makefile.template order) - use ports-gcc on gcc archs like other C++ ports (here a C++11 compiler is required) Looks good.
git-crypt.tgz
Description: Binary data
-- jca | PGP : 0x1524E7EE / 5135 92C1 AD36 5293 2BDF DDCC 0DFA 74AE 1524 E7EE