[lmi] Directory structure for multiple-architecture builds

[Greg Chicares]

Vadim--I'm hoping to build lmi and its dependencies simultaneously for
various hosts, all in the same chroot (running msw binaries via wine):
  - x86_64-linux-gnu (gtk)
  - x86_64-w64-mingw32
  - i686-w64-mingw32 (until we can convert users to x86_64)
and ultimately with clang as well as gcc, and I'm wondering what
directory structure to use. I imagine this is a problem you've already
solved; do you have a recommendation? I see (at least) three options:

(1) All binaries in one directory, with host-prefixed names, like gcc:
  i686-w64-mingw32-lmi
  x86_64-w64-mingw32-lmi
  x86_64-linux-gnu-lmi
I've never seen this style for anything but cross compilers. I could do
this for lmi itself, but dependencies like libxml2 don't generally have
prefixed names, and I'm looking for a single uniform approach.

(2) Host-prefixed subdirectories, like debian multiarch:
  /usr/lib/i386-linux-gnu/libxml2.so.2
  /usr/lib/x86_64-linux-gnu/libxml2.so.2
For msw builds, libxml2 installs binaries in two places:
  /opt/lmi/local/bin/libxml2-2.dll
  /opt/lmi/local/lib/libxml2.dll.a
and I could insert an autotools triplet before or after bin/ and lib/ ,
or I could append it to the string "local", thus:
  /opt/lmi/local-i686-w64/bin/libxml2-2.dll
  /opt/lmi/local-x86_64-w64/bin/libxml2-2.dll
(i.e., build libxml2 with 'prefix=/opt/lmi/local-x86_64-w64').

(3) A single host-prefixed directory for each architecture (and even
for each compiler). I suppose this is what I'd get by specifying
something like 'exec-prefix=/opt/lmi/i686-w64'. Then we'd have
  /opt/lmi/src [same for all]
  /opt/lmi/i686-w64/bin  [lmi's own binaries]
  /opt/lmi/i686-w64/local/bin/libxml2-2.dll
  /opt/lmi/i686-w64/local/lib/libxml2.dll.a
  /opt/lmi/x86_64-w64/bin [lmi's own binaries]
  /opt/lmi/x86_64-w64/local/bin/libxml2-2.dll
  /opt/lmi/x86_64-w64/local/bin/libxml2.dll.a

(1) doesn't seem like a good choice. I'm not sure whether there's
really much difference between (2) and (3), but (3) seems to maintain
a more obvious separation. What do you think?

Whatever choice we make, it might make a lot of sense to use an empty
string for the triplet distributed to end users, so that Kim can just
build a single production version as usual.