[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: ARMv4-support in armel/squeeze?

+++ Luke Kenneth Casson Leighton [2010-12-19 21:04 +0000]:

>  weelll... how about creating an easy means for anybody to create
> their _own_ debootstrap'd cross-compiled starting point, based on
> _their_ decisions and requirements, and debian can host the most
> popular ones?

I quite agree (and have done for years), and am actually working on it
at the moment. 

>  by "easy" that would mean "one config file" followed by "one command"
> (followed by an enormous wait and having large amount of hard drive
> space chewed up but that's only to be expected).

That would be the goal. 

> why is this information not being recorded
> into a reproducible form whereby it is a simple matter of "one config
> file" followed by "one command" (followed by an enormous long wait) ?

Becauswe there hasn't been a good place to write it down in Debian's
recursive-native-build view of the world. 

>  is it somehow... _deliberate_ that nobody has created such an
> automated debootstrap creater?  there's been 14 debian ports so far:
> _surely_ it occurred to someone that it would be a good idea to
> automate the process of creating new ports?  

Of course it occurred. The problem has been obvious for years, but
it's not simple to solve in a repeatable manner. I believe we do now
have sufficient momentem and interest to make this happen (in Ubuntu
first, I suspect, but if it's done right then it's all portable, and
it's not worth doing unless it's done right)

> is there a debian policy
> which _prevents_ people from creating an automated debootstrap
> creater? 

Of course not, but the fact that cross-building was not viewed as
important meant an ad-hoc solution was always expedient.

> (i'm aware that all ports must be native-compiled but that
> simply isn't possible for the "initial" port, so you need some sort of
> 3-stage-gcc-like-bootstrap "thing" to get to that point).

Indeed. And the gcc/glibc-bootstrap has now been incorporated into the
cross-gcc packaging. A load more packages need similar work doing (at
least one package in every circular build-dependency loop).

I am documenting my current thinking (which has been developed with
the help of others - I'm not that clever on my own) on the
bootstrapping issue here: http://wiki.debian.org/EmdebianBootstrap
Comments welcome as there are almost certainly holes in it (there is
more detail to come).

I'd love to hear from Konstantinos what his list of problem packages
is, and what hackery was done to get each initial build done. 

Principal hats:  Linaro, Emdebian, Wookware, Balloonboard, ARM

Reply to: