Hi all,
I wanted to hear your opinions about how we present commands in the cross-lfs book. When we first started writing the book, we decided it would be easier to use variables for the target triplets to be used throughout the cross-tools and temp-system section.
Something like this:
LFS_HOST=i686-pc-linux-gnu LFS_TARGET=sparc64-unknown-linux-gnu
Then in the instructions in the book we use those variables for our various --build, --host and --target flags, for example, --target=${LFS_TARGET}. This seems all well and good for it helps generalize the instructions for a multi-arch book.
However, the question comes up, to what extent are variables helpful in the book? For editing purposes, it becomes much easier to use them in place of other common commands to be run. But does this obscure information from the reader?
Take a look at this example and note the configure command:
http://linuxfromscratch.org/~jhuntwork/cross-lfs/x86_64/cross-tools/glibc-startfiles.html
Compare that page with this one:
http://linuxfromscratch.org/~jhuntwork/cross-lfs/sparc64/cross-tools/glibc-startfiles.html
Notice the use of $(BUILD32) in the x86_64 book that would contain (at least) the -m32 tag appended to the CC variable. It was thought that if we could have the user define the gcc tags needed to build for 32 or 64 in the book for each arch, we could generalize the pages more and keep it less arch specific. How does the community feel about this?
-- Jeremy Huntwork -- http://linuxfromscratch.org/mailman/listinfo/lfs-dev FAQ: http://www.linuxfromscratch.org/faq/ Unsubscribe: See the above information page
