lmi-commits
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[lmi-commits] [lmi] odd/multiarch 993027b 4/5: Document an idea that is


From: Greg Chicares
Subject: [lmi-commits] [lmi] odd/multiarch 993027b 4/5: Document an idea that is less good than was hoped
Date: Wed, 1 May 2019 17:30:51 -0400 (EDT)

branch: odd/multiarch
commit 993027b69dc6d9350604f80d11b55f37784c3f4a
Author: Gregory W. Chicares <address@hidden>
Commit: Gregory W. Chicares <address@hidden>

    Document an idea that is less good than was hoped
---
 set_arch.sh | 29 +++++++++++++++++++++++++++++
 1 file changed, 29 insertions(+)

diff --git a/set_arch.sh b/set_arch.sh
index 67ff3ee..f9707b5 100755
--- a/set_arch.sh
+++ b/set_arch.sh
@@ -46,6 +46,35 @@
 #   ln --symbolic --force --no-dereference 
/opt/lmi/"${LMI_COMPILER}_${LMI_TRIPLET}"/bin /opt/lmi/bin
 # and likewise for /opt/lmi/local etc.
 
+# To cross-build autotoolized libraries, configure with:
+#   --host=$LMI_TRIPLET
+#   --exec_prefix=$prefix/$LMI_TOOLCHAIN
+# where
+#   LMI_TOOLCHAIN=${LMI_COMPILER}_${LMI_TRIPLET}
+#
+# Supported values:
+#   LMI_COMPILER : gcc, clang
+#   LMI_TRIPLET  : i686-w64-mingw32, x86_64-w64-mingw32, x86_64-pc-linux-gnu
+#
+# Examples:
+#
+#  LMI_TOOLCHAIN  LMI_COMPILER  ----------LMI_TRIPLET----------
+#                               cpu     -vendor -[kernel-]system
+#
+#      gcc_msw32       gcc      i686    -w64            -mingw32
+#      gcc_msw64       gcc      x86_64  -w64            -mingw32
+#    clang_gnu64      clang     x86_64  -pc     -linux  -gnu
+#
+# Rationale: Triplets are required for selecting cross compilers. GNU
+# triplets are in widespread use, although debian uses its own:
+#   https://wiki.debian.org/Multiarch/Tuples#Why_not_use_GNU_triplets.3F
+# The main reason for lmi to establish its own is that the official
+# triplets for msw are so poor. For example, "i686-w64-mingw32" has
+# substrings "64" and "32". Both seem to suggest the CPU's word size
+# (contradicting each other), but actually neither does--the word size
+# is determined by the "i686" field. See:
+#   https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=622276
+
 foo()
 {
 local   lmi_build_type



reply via email to

[Prev in Thread] Current Thread [Next in Thread]