On Thu, Nov 05, 2009 at 09:08:28PM +0100, Jann Traschewski wrote:
I wrote:
- Determine the endianness by used crosscompiler (I put "bird_cv_c_endian=little-endian" for my Mipsel target into CONFIGURE_ARGS as a workaround). Maybe this check could be done using the configure-script.
After the release of V1.1.5 I was able to crosscompile BIRD using the OpenWRT buildsystem without my "hack".
Yes, i started to work on proper BIRD package for OpenWRT (based on your Makefile) and fixed this issue in upstream. Building just IPv4 version is easy, building both IPv4 and IPv6 version in two packages from one Makefile is a little tricky :-) .
- Find out how to get a package compiled with the buildsystem depending on "readline" (I put "--disable-client" into CONFIGURE_ARGS as a workaround). I managed to compile it manually by crosscompiling the readline lib before.
Further I figured out how to handle the readline dependency.
Probably ncurses have to be handled in the same way.
Looking closer into the buildsystem I still have some questions: - It might be worth to split "bird" and "birdc" into two packages since birdc depends on libncurses and libreadline which takes much space on a router. What do you think?
I think that bird without birdc is not really useful, but probably it might be useful to compile birdc without libncurses and libreadline (or perhaps just without libreadline).
- Do I have to compile "birdc6 (or bird6c?)" if I want to provide the "bird6"-package or is the client the same for IPv4 and IPv6?
It is expected to compile birdc6. The command set might be different (but probably only difference is an absence of OSPF in birdc6). Also default socket name is different, but this could be easily changed. -- Elen sila lumenn' omentielvo Ondrej 'SanTiago' Zajicek (email: santiago@crfreenet.org) OpenPGP encrypted e-mails preferred (KeyID 0x11DEADC3, wwwkeys.pgp.net) "To err is human -- to blame it on a computer is even more so."