Date: Sun, 18 Nov 2018 18:39:44 +0100 From: Harry Schmalzbauer <freebsd@omnilan.de> To: ports@freebsd.org Subject: building mail/sendmail depends on 191 ports like python2, python3, llvm6 (while same in base) etc... Message-ID: <0de7a35d-559a-e17e-7d91-76b749182766@omnilan.de>
next in thread | raw e-mail | index | archive | help
Hello, mail/sendmail/Makefile reads USES= cpe uidfix groff This leads to a dependency of 191 ports to build sendmail. Including two versions of python framework, a compiler, which corresponds to the one provided by base, and things like wayland, cups, w3m, git etc.... (and default options are highly inconsistent for IDN e.g in this dependency mess). I have no idea why https://svnweb.freebsd.org/ports?view=revision&revision=460448 was needed, so where to start fixing that? Reverting that commit (removing groff from USES) reduces dependency by 183 ports (8 instead of 191). I'd love to see porters handbook showing samples of reusable snippets, which maintainers can think about and adapt before incorporating it and still save a lot of time, instead of incorporationg completely non-understandable/-documented makros, spending hours for reading what's the idea behind USES+= groff afterwards. Somebody is forced to do the latter :-( Btw, SMTPUTF8 option leads to domain.c:17:10: fatal error: 'unicode/uidna.h' file not found #include <unicode/uidna.h> ^~~~~~~~~~~~~~~~~ If I wouldn't feel like I don't understand anything from ports anymore, I'd love to help out with patches, but I don't know where to start and I simply don't get the idea of the decisions for the way FreeBSD's ports tree was pushed to. Just wanted to point out this circumstance and tell anybody else, wondering how things can go so wrong, to pay attention to USES lines in port's Makefile. Thanks to everybody for keeping the ports tree alive and healthy from the users point of view. -harry
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?0de7a35d-559a-e17e-7d91-76b749182766>