[CentOS-devel] Centos 7.i686

Thu Jan 8 19:51:17 UTC 2015
Johnny Hughes <johnny at centos.org>

On 01/05/2015 02:32 PM, Johnny Hughes wrote:
> On 01/05/2015 07:44 AM, Fabian Arrotin wrote:
>> On 05/01/15 14:37, Johnny Hughes wrote:
>>> On 01/05/2015 07:01 AM, Fabian Arrotin wrote:
>>>> On 05/01/15 13:51, Johnny Hughes wrote:
>>>>> On 12/26/2014 05:24 AM, Karanbir Singh wrote:
>>>>>> On 25/12/14 16:54, Bob Lightfoot wrote:
>>>>>>> Dear Devs: I have been reading this mailing list since C7
>>>>>>> came out and of late have not seen much progress on C7.i686
>>>>>>> aka 32 bit.  I don't see a SIG for it on the Centos Pages.
>>>>>>> I am wondering has the idea been dropped or where does it
>>>>>>> stand?
>>>>>>
>>>>>> the larger goal is to let this effort be user-led, so
>>>>>> starting up a SIG might be a good way to go here. I know that
>>>>>> the original bootstrap had most of the builds done - and
>>>>>> Andreas ( assited by others ) had gotten most of the bits
>>>>>> done. There would be a need to run the updates, and then also
>>>>>> identify what portion of the distro is not going to make it
>>>>>> to i686 at all. The rest from there should be easy...
>>>>>>
>>>>>> As I did with the powerpc effort, happy to host a i686
>>>>>> specific google hangout where I can walk people through the
>>>>>> build process, and what they need to do in order to affect
>>>>>> builds in the centos buildsystem.
>>>>>>
>>>>>> - KB
>>>>>>
>>>>
>>>>> I now have a working syslinux and kernel in git.centos.org
>>>>> under c7-i686
>>>>
>>>>> I am building those every time they update for i686 as well as 
>>>>> x86_64
>>>>
>>>>> I have all the RPMs currently built, including the 
>>>>> java-1.6.0-openjdk built that was an issue .. I'll post the
>>>>> RPM list and what is missing (compared to x86_64) and we can
>>>>> try to figure out what we need to make build (We may need to
>>>>> change some other things that they made exclusivearch x86_64,
>>>>> like they did syslinux)
>>>>
>>>>> Then we can get an i686 test spin out.
>>>>
>>>>> Here is the kernel and syslinux links for i686
>>>>
>>>>> https://git.centos.org/log/rpms!kernel.git/refs!heads!c7-i686
>>>>
>>>>> https://git.centos.org/log/rpms!syslinux/refs!heads!c7-i686
>>>>
>>>>
>>>>
>>>> I had to reinstall an old thinkpad (family laptop used by the
>>>> kids) and wanted to build a C7/i386 livecd to install it with
>>>> minimal desktop, but some packages (like ibus-sayura) are missing
>>>> from buildlogs.centos.org and no build logs either for i386,
>>>> meaning no built was even tried .. Can we just massively try to
>>>> build all packages to i386 to at least have logs and see why they
>>>> fail (or not) ?
>>
>>> c7.00.02/ibus-sayura/20140529190519/1.3.2-3.el7.i386/
>>
>>
>> Yeah Johnny :-)
>>
>> I saw that one too, but what I meant was that not all of those
>> packages were tried again after that, and were depending on other
>> packages.
>> In that specific example (ibus-sayura) the
>> http://buildlogs.centos.org/c7.00.02/ibus-sayura/20140529190519/1.3.2-3.el7.i386/root.log
>> clearly mentions a need for pyOpenSSL, which itself had been built
>> after
>> (http://buildlogs.centos.org/c7.00.02/pyOpenSSL/20140529192726/0.13.1-3.el7.i386/)
>> so when I said 'massive rebuild' I meant retrying all failed packages
>> after the first run and that would (probably) succeed now that build
>> deps are satisfied :-)
> 
> Right .. I did that once already, but obviously I missed a few.
> 
> I am creating that list of things missing right now, once I get it, I
> will look at the build logs to see what we need to try to rebuild.

Here is a list of Packages that have a .x86_64.rpm but will not have an
i686.rpm

If any of these are a show stopper for someone, we will need a way to
make them build:

http://fpaste.org/167305/

Hopefully we will have an installable test tree soon.


-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 198 bytes
Desc: OpenPGP digital signature
URL: <http://lists.centos.org/pipermail/centos-devel/attachments/20150108/2066840e/attachment-0008.sig>