Re: [PATCH 3/3] {e}glibc: Fix various multilib build bugs

[Ray Donnelly <mingw dot android at gmail dot com>]

I'm not having the best time of it today! Bryan, I forgot to CC the ML
(so most of this you got already) but also, there's a dumb error in my
patch:

target = $(get_multilib_target "${extra_flags}")

needs to be changed to:

target=$(get_multilib_target "${extra_flags}")

On Tue, Feb 4, 2014 at 8:14 PM, Bryan Hundven <bryanhundven@gmail.com> wrote:
> Ray, Cody, list,
>
> On Tue, Feb 4, 2014 at 10:49 AM, Ray Donnelly <mingw.android@gmail.com> wrote:
>> Here is the new version using Cody's get_multilib_target () function.
>>
>> Cheers,
>>
>> Ray.
>
> I have some comments on this commit, but it's attached :(
> I recommend using the patchbomb method:
> http://crosstool-ng.org/hg/crosstool-ng/file/529a71ea091e/docs/C%20-%20Misc.%20tutorials.txt#l183
>

Sorry about the attachments, I've only submitted a few minor patches
so far and nothing that needed much discussion.
Is patchbomb easy to setup with gmail? If you can point me at some
instructions that'd be appreciated.
IMHO ctng should move to Git and use Gerrit (or failing that maybe
just github) for these code reviews, but I understand that this would
be quite a significant time drain to implement.

> Anyways, it seems to me that get_multilib_target should be stubbed in
> scripts/build/arch/multilib.sh, and implemented in the respective
> scripts/build/arch/<architecture>.sh
> Source multilib.sh after this line:
> http://crosstool-ng.org/hg/crosstool-ng/file/529a71ea091e/scripts/crosstool-NG.sh.in#l140
>

I agree we could move get_multilib_target somewhere more general
purpose, but I'd put it into multilib.sh and leave it at that.
Personally, I prefer having the full gory details available in (and
easily testable from) a single function rather than splitting it up.

> That way, any package - beyond (e)glibc - can obtain the multilib
> target info if needed.
>
> Other points are easier to discuss inline. Cody, Ray; what are your thoughts?
>
> -Bryan

--
For unsubscribe information see http://sourceware.org/lists.html#faq