[Clfs-dev] Latest build updates(gcc 4.8.2 etc) and config.cache changes

[Martin Ward]

On 06/11/13 16:04, William Harrington wrote:
>
> On Nov 5, 2013, at 3:15 PM, Martin Ward wrote:
>
>> checked out a full build using 32 bit kernel 2.6.30.5, glibc 2.9, and 
>> gcc 4.4.1
>> pure 64 build
>>
>> used kernel headers from 3.12
>> gcc-4.8.2, no updates as prefer unpatched version
>> eglibc 2.18 r24452 (much the same as the book version, poss. a couple 
>> of nptl update for ppc)
>>
>
> I have 24420 when I built with the dual p3 last week. May wait till 
> 2.19 before anymore updates.
>
>> mpfr 3.12 has small amount of bugs, arch has a patch for them
>>
>
> Yeah we'll need to include them.
>
>> binutils-2.23.2
>> added --enable-plugins
>> --enable-threads, not strictly necessary but added anyway
>>
>
> Is that for final system?
i add that to both, can't remember where i saw why, will try to find 
refereence later
>
>> config.cache changes
>> Coreutils-8.21 guessed yes
>> Findutils-4.4.2 guessed yes
>> Flex-2.5.37 guessed yes
>> Gettext-0.18.3.1 guessed yes
>> Grep-2.14 set to yes, so we can remove this?
>> M4-1.4.17 guessed yes except gl_cv_func_mbrtowc_null_arg which as 
>> been replaced by gl_cv_func_mbrtowc_null_arg1 and 
>> gl_cv_func_mbrtowc_null_arg2 both guessed yes
>> Tar-1.27 all guessed yes except gl_cv_func_mbrtowc_null_arg which as 
>> been replaced by gl_cv_func_mbrtowc_null_arg1 and 
>> gl_cv_func_mbrtowc_null_arg2 both guessed yes and 
>> ac_cv_func_realloc_0_nonnull not found in configure at all
>>
>
> I went through the config caches a while back. We have grep 2.15 now, 
> though.
ooh missed that .., will check later on that
>
> I'll double check them all but I saw where there were yes being 
> guessed multiple times. I prefer to not have them guessed.
> The invalid variables we can take out or change to the proper 
> variables for null and any others.
>
>> Texinfo-5.2 no need for separate make steps, will go straight through 
>> a make to completion
>>
>
> Built with Texinfo 5.2. We'll need to add sed commands to binutils for 
> final system as we do in cross tools and temp system.
or a patch
>
>> on to boot method later
>>
>>
>
> As far as check. We need to move it to tools. If someone doesn't build 
> the test suite tools, check won't get installed and kbd will fail in 
> final-system.
i think that is what i did last build
> Sure, someone can go ahead and install check even not the rest of the 
> test suite commands, but the user has no clue that kbd will require 
> check until they go to build it, especially if they aren't familiar 
> with the build process.
agreed
>
> As far as static libraries, shall we remove them for tools and 
> final-system, too? In my last build I built a shared library for flex 
> and bison, as HLFS does. I've removed all static libs. I saw LFS doing 
> it and gave it a whirl, although they still install the static bison 
> and flex libraries.
to be honest, these days, disk space is not an issue, i'd rather keep 
them , just in case something happens later on, Only need to remove if 
you are a purist IMHO
>
> I also used automake 1.14 in my current build, as well. Now that Linux 
> 3.12 is out, may bump up to that.
will check that out as well
>
> Thank you for your updated progress report.
>
no problem
>> regards
>>
>> Martin
>> _______________________________________________
>> Clfs-dev mailing list
>> Clfs-dev at lists.cross-lfs.org
>> http://lists.cross-lfs.org/listinfo.cgi/clfs-dev-cross-lfs.org
>
> Sincerely,
>
> William Harrington
> _______________________________________________
> 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