Matthew Burgess wrote:
Andrew Fyfe wrote:
Greg Schafer (http://www.diy-linux.org/x86-reference-build/) has been building pass 1 of binutils and gcc dynamically for a while without any problems.
Thanks. Shockingly, I actually read the book for once, to see what it had to say on the matter (5.3.1):
"However, it is worth noting that an overall successful LFS build can still be achieved when the first two packages are built dynamically."
So, it would seem from general experience this is quite safe to do, and the slight element of independence from the host that the static build gives us won't be noticed. Does this mean that all of section 5.3.1 can go now, as it only seems to be there to introduce the explanation of why we statically link the pass 1 builds in chapter 5?
Matt,
Our solution in BLFS is generally to insert a note telling the problem and workaround, but leaving the preferred approach in the mainline instructions.
Right, but like I said above, I'd rather provide just one way for everyone to follow. I realise that BLFS is more about picking and choosing things to build and ways of building, but as you'll no doubt be aware, we try and minimise such choices in LFS. If building dynamically works for everyone, I'd rather just have everyone use that - especially as it concerns the toolchain which is easy enough to screw up as it is :)
OTOH, I am concerned that LFS 6.0 is broken and strongly encourage a 6.0.1 bugfix release. This is something that even Linus did with 2.6.8.1.
And something I had to do for 5.1.1 :(
This issue has been with us for at least 5 months though - it first reared its head in November last year I believe. Therefore, is there a real urgency about fixing things up in an official release? I understand folks concerns about getting a release out to address this, but to be honest we've had few reports of hosts unable to build LFS-6.0 due to this issue. Until we release a new LFS, folks won't realise that LFS-6.0 is broken.
Don't get me wrong, I don't mind Gerard doing a release ( :) ) if that's what is felt is the best approach. I just wonder if it's worth basing the release off of 6.0 (i.e. no package upgrades, etc.) or if we branch from trunk/ and freeze for release once the trivial fix goes in. Randy and others seem to think it's pretty stable already, so getting it out the door shouldn't take too long.
Regards,
Matt. -- http://linuxfromscratch.org/mailman/listinfo/lfs-dev FAQ: http://www.linuxfromscratch.org/faq/ Unsubscribe: See the above information page
