Autotools

Alexander E. Patrakov patrakov at ums.usu.ru
Thu Dec 7 04:36:45 PST 2006


Randy McMurchy wrote:
> Alexander E. Patrakov wrote these words on 12/07/06 00:10 CST:
>   
>> This is a formal request to either integrate Tushar's hint on installing 
>> multiple versions of autotools into the LFS book, or move autotools to 
>> BLFS and integrate the hint there.
>>     
>
> -1. I can't see multiple versions of autotools being installed, be
> it LFS or BLFS. But, if it has to be, then it would belong in LFS.
> At least that's the way I see it.
>   
Point taken.
> Popt is not an issue, or at least I'm not aware of any issue. Is
> there some patch that needs to be applied that is not in the book
> yet? If not, Alexander's message is borderline FUD.
>   
As for popt, from the existing BLFS standpoint, my message is indeed 
borderline FUD.
> As far as Reiser support, this seems to be an issue specific to
> Reiser and should be addressed on the Reiser page. Right?
>   
Yes, except that there is currently no reiser4 page, but it is not my 
point (I could just as well create it in the wiki). The question is 
whether we install /usr/bin/auto* just for beauty (with no intention to 
call it), or for a functional purpose. As far as I can remember the past 
discussions, autotools are a part of the development platform that LFS 
aims to be.

New autotools don't make much harm if one follows BLFS to the letter, 
exactly because BLFS usually doesn't call them.

New autotools certainly serve a good job for starting new applications 
and are the only way forward. I.e., the developer writes configure.in 
and Makefile.am for his own application in LFS, and everything works 
just fine.

However, a "development platform" is not just about starting new 
applications, it's also about participating in various existing projects 
and applying beyond-BLFS modifications to LFS and BLFS software. And the 
problem is that, out of 7 packages that were re-autotooled in the 6.2 
branch of the LiveCD, 4 broke because of the upgrade of autotools. For 
me, that's too much.

I understand that, in order to see the breakage, one has to deviate from 
the books. I also see that the books don't use autotools much, and 
(correctly) rely on pre-generated ./configure scripts where possible. 
The question is whether the "real" use of autotools in existing projects 
is just assumed to never happen.

-- 
Alexander E. Patrakov



More information about the lfs-dev mailing list