[Clfs-dev] Latest CLFS Build Notes Pure64 08-08-2013

[Martin Ward]

On 12/08/13 18:23, William Harrington wrote:
>
> On Aug 12, 2013, at 9:34 AM, Martin Ward wrote:
>
>> Hi
>>
>>
>> GCC-4.8.1
>> really think we should disable the fix-includes script in both temp 
>> and final systems see here for rational http://ewontfix.com/12/
>>
>> sed -i 's@\./fixinc\.sh at -c true@' gcc/Makefile.in does the job(copied 
>> from Arch Linux)
>
> From the looks of it, that would be good.
>
> As far as my issue with -Wno-narrowing and the other errors, it seems 
> to be from hosts that are using gcc < 4.7. Have you used any hosts 
> which use gcc  < 4.7?
>

Yep mine usual host is 32 bit CLFS build with kernel 2.6.30.5, glibc 
2.9, gcc 4.4.1, which incidentally was cross-compiled from an LFS live CD
>> Installing linux headers
>> In both  temp system and final the headers are installed direct in to 
>> their final locations, I prefer cleaning the directories beforehand 
>> rather than after, Personal choice though.
>>
>
> Cross-tools there is nothing in /tools/include at all before the install.
>
> in the final-system when the headers are installed, if folllowing the 
> boot method, eudev headers will be in /usr/include. WIth the chroot 
> method, there should be nothing there.
>
>
Suppose its a personal choice
>> Section 7.5
>> In the boot phase shadow configuring has a --config.cache option when 
>> it's not needed anymore
>>
>
> Finally fixed.
>
>> 7.7.2 and 10.58.2 Sysvint
>>
>> Perhaps we should add a --noclear option to the first agetty line, 
>> this enables the user to see boot messages rather the been cleared on 
>> the first boot
>>
>
> I usually do that myself. We could add it or insert a note about it.

sounds reasonable
>
>>
>> 10.27.1
>>
>> Bison 2.7.1, still need ac_cv_prog_lex_is_flex=yes on the command 
>> line other wise it won't configure
>> Bison 3.0 is even worse, not only that, couldn't get it to 
>> cross-compile, seems to run an architecture dependant to get generate 
>> doc/help files, but also needs ac_cv_prog_lex_is_flex=yes to succeed, 
>> but complains about a hard coded m4 invocation(cross-tools) from 
>> flex(installed in tools) in generating some test files
>>
>
> I didn't bother with Bison 3.0 as it was compiling a host binary with 
> the target compiler so the host couldn't run the binary to continue 
> the build.
>
> I removed ac_cv_prog_lex_is_flex as it was being configured properly 
> in the last 10 builds I've done. I didn't have an issue going from 
> x86_64 to ppc either.
>
> I don't know why that is cropping up sometimes. I've been using Debian 
> Jessie as a host and the updated lfs livecd.
>
weird, wonder what it is ? See my next post about bison, flex and m4
>>
>> cheers
>>
>> martin
> _______________________________________________
> Clfs-dev mailing list
> Clfs-dev at lists.cross-lfs.org
> http://lists.cross-lfs.org/listinfo.cgi/clfs-dev-cross-lfs.org
>
regards

Martin