public inbox for gentoo-dev@lists.gentoo.org
 help / color / mirror / Atom feed
From: Debian User Jean-Baptiste Note <noteje@poly.polytechnique.fr>
To: Chouser <chouser@gentoo.org>
Cc: gentoo-dev@gentoo.org
Subject: [gentoo-dev] gentoo ppc
Date: Tue, 8 Jan 2002 16:45:57 +0100	[thread overview]
Message-ID: <20020108164556.A1605@debian> (raw)
In-Reply-To: <20020107190114.C377053@plato.zk3.dec.com>

Hello

I'm glad to announce that emerge system works for (almost)all packageson ppc.
>From these I have to substract the x86-specific packages. and iproute2,
that gives me some problems maybe because of -Werror only.
I have kept good record of the steps i took.

Just to introduce myself, i'm 22 years old, a student in a French engeneering
school. Although I have been using a computer for some time now, and debian linux
for a little more than one year, i must say i'm no linux expert.
I have very little experience in development for instance, but i'm willing
and hopefully able to learn fast :)

>From this comes two points :
1/ gentoo is extremely flexible. Port isnt hard, or i wouldn't have succeeded
in advancing so much. I never delved into portage, and barely edited some .ebuilds files.
I have gathered enough information so that i can write a gentoo-porting howto
from a working debian install, that can work as a base for an alpha port,
for instance, that should not be hard either.
Thank you for the great job of Gentoo in the first place !

2/ I am willing to do some modifications to portage to allow more cross-platform
flexibility. But, it obviously needs to be concerted.
So i'm gonna tell you boringly of the problems i encountered and some solutions
that i though of.

Problems are
a) platform-specific patches (gcc, patch itself)
b) platform-specific options (for instance, a `use altivec` one for
powerpc processors)
b) platform-specific packages (bootloaders for instance)

b)
I think portage needs a little tweaking i'm willing to do.
this would go thus (let's take the function src_uncompress as an exemple)
* look in the .ebuild for a platform-specific function to be executed
if it exists. 
* if src_uncompress_ppc exists on my ppc machine,
then execute it regardless of the existence of src_uncompress
* if no such platform-specific function is found, then
fallback on scr_uncompress.

This would allow a smooth transition :
--existing .ebuild files still work, and a default behaviour
is given to the platform-specific packagers
--you can still "factorize" platform-independent instructions
in the scr_uncompress function, and call it in the platform-specific
function
--ppc/alpha developers can safely work on their specific part
without endangering the x86 part

It would be nice, too, if some ${ARCH} variable computed by portage
was available to the .ebuild writer.

a)
The problem is that you don't necessarily want the x86 user to download
a ppc-spceific patch.
then you could define a SCR_URI_X86 SRC_URI_PPC additional to the already
existant SRC_URI if the ${ARCH} is matched. Better hide it in portage
than in a ugly if statement in the .ebuild file, i think (once again,
code factorization)

c)
I should think that platform-dependent utilities are not so much needed.
well, apart from x86 assembly compiler and bootloaders, that is :)
the answer, generally speaking, is virtual packages, if all platform can
provide their version of one tool. (ybin as bootloader for instance)
this is certainly the most difficult part, and i haven't given it much
though.

One more personal remark, not to bother you with my life, but :
i'm still a student, and do not have a hell of time :(
that'll be worse even after i suppose anyways.
Things i want to do
*solve the platform-specific-packages design problem
*write a gentoo-bootstrap-from-debian howto to help support other architectures (alpha ?)
*package ybin ? not really necessary, since most will dual-boot or
work chrooted
*hack portage. i'm learing python (not so tough :) and will look into
portage. but to implement, things need to be settled.

I'm done. I didn't know how to introduce myself, if i may sound awkward or
anything, please do not heed.
(My mastering of the English language is not good, either).

Thank you for your time, Jean-Baptiste Note.

-- 
I'm an occurence of the I love GNU virus.
Please help me spread.


       reply	other threads:[~2002-01-08 15:49 UTC|newest]

Thread overview: 12+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
     [not found] <20020107190114.C377053@plato.zk3.dec.com>
2002-01-08 15:45 ` Debian User Jean-Baptiste Note [this message]
2002-03-25 12:34 [gentoo-dev] gentoo ppc Pieter Van den Abeele
2002-03-25 13:03 ` Kain
2002-03-25 13:59   ` Pieter Van den Abeele
2002-03-29  8:53     ` Kain
2002-03-29  9:24       ` Pieter Van Den Abeele
2002-03-30  2:57         ` Kain
2002-03-30 19:01           ` Pieter Van den Abeele
2002-03-31  3:42             ` Fuper
2002-04-01 12:12               ` Pieter Van den Abeele
2002-04-01 14:29                 ` Pieter Van den Abeele
2002-04-04 11:14                   ` linux-dev

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20020108164556.A1605@debian \
    --to=noteje@poly.polytechnique.fr \
    --cc=chouser@gentoo.org \
    --cc=gentoo-dev@gentoo.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox