[MacPorts] #35217: boost @1.50.0 Porfile does not pass configure.c{xx}flags to boost build
MacPorts
noreply at macports.org
Thu Jul 19 14:25:57 PDT 2012
#35217: boost @1.50.0 Porfile does not pass configure.c{xx}flags to boost build
----------------------------------------+-----------------------------------
Reporter: andrew.c.morrow@… | Owner: adfernandes@…
Type: defect | Status: closed
Priority: Normal | Milestone:
Component: ports | Version: 2.1.1
Resolution: wontfix | Keywords:
Port: boost |
----------------------------------------+-----------------------------------
Changes (by adfernandes@…):
* status: new => closed
* resolution: => wontfix
Comment:
I agree, this sounds like a great suggestion.
Unfortunately, the boost documentation lies... a lot.
Disclaimer: I've done a lot of (non-MacPorts) work building `boost` for
iOS, and it really is not fun at all.
The main problem with adding compiler flags like you show is that it only
sometimes works. If you look at the `darwin.jam` file (and if you can
manage to read it, since virtually all of it is undocumented, untested,
not-working, and so on) you'll see that the build system magically adds
and removes compiler flags depending on what it thinks you want to do. It
modifies the flags based on compiler, SDK, operating system, version, and
so on. As far as I can tell, none of it is really tested except for what
is explicitly stated on the boost documentation page.
You'll note that as of `boost-1.50.0` you cannot use `apple-clang++-3.1`
to build `boost`, even though clang-2.8 and later are technically
supported.
You know, I've actually looked a lot into this because of #35172. What I
wanted to do was build boost with gcc-4.7 and link the gcc-runtime in
statically. That would entail adding `-static-gcc` and `-static-c++`
flags. However, the `boost` build system won't let you do that. I don't
know why. It just has a snarky "man, that's a bad idea, so I won't let you
do that" message in the jam file. (I suspect it has to do with
[http://stackoverflow.com/questions/3430400/linux-static-linking-is-dead
this linux-ism], but do not know.)
My feelings are that unless this provides a reasonable benefit to a
reasonable subset of users, it risks too much breakage to do this.
--
Ticket URL: <https://trac.macports.org/ticket/35217#comment:3>
MacPorts <http://www.macports.org/>
Ports system for Mac OS
More information about the macports-tickets
mailing list