[OE-core] [PATCH] boost: add python to default PACKAGECONFIG options

[Huang, Jie (Jackie)]

> -----Original Message-----
> From: MacLeod, Randy
> Sent: Wednesday, August 23, 2017 09:27
> To: Richard Purdie; Andre McCurdy; Huang, Jie (Jackie)
> Cc: OE Core mailing list
> Subject: Re: [OE-core] [PATCH] boost: add python to default PACKAGECONFIG
> options
> 
> On 2017-08-22 03:55 PM, Richard Purdie wrote:
> > On Tue, 2017-08-22 at 12:51 -0700, Andre McCurdy wrote:
> >> On Tue, Aug 22, 2017 at 12:42 AM,  <jackie.huang at windriver.com>
> >> wrote:
> >>>
> >>> From: Jackie Huang <jackie.huang at windriver.com>
> >>>
> >>> We want to provide python libs by default, and some other
> >>> popular Linux distributions like redhat/fedora does the same.
> >> Has something changed? Is there anything in oe-core or meta-oe which
> >> now needs boost python support enabled?
> >
> > Does it even actually build on all arches? >
> > I'm rather behind with patch review right now as people are putting so
> > many patches out there with what seems like minimal testing and
> > expecting me/Ross and the autobuilder to figure out the problems for
> > them.
> >
> > Each time one fails, it blocks the queue with the rest of them in and I
> > have to retest until I get something clean.
> >
> > Testing this change is quite a way down my priority list I'm afraid.
> >
> 
> Maybe this is better done in our distro configuration.
> 
> We're trying to minimize differences with oe-core and
> the boost python support doesn't seem to be much of an
> impact based on looking at logs rather than a having a
> detailed understanding of boost.
> 
> We've had this commit in our local branch since before:
> 
> CommitDate: Wed Dec 21 22:16:31 2016 -0800
> 
>      boost: add python to default PACKAGECONFIG options
> 
>      We want to provide python libs by default, and some other
>      popular Linux distributions like redhat/fedora does the same.
> 
>      (LOCAL REV: NOT UPSTREAM) -- sent to oe-core on 20160930
> 
> so it certainly builds for all qemu* for glibc configurations.
> 
> I'll see if we can add musl builds to our automated builds tonight.
> 
> Jackie,
> 
> Did/Can you build for musl?

No, I didn't build for musl yet, but yes, I can and will do the test with musl
and fix issues as needed.

> Do you have any idea of the impact of this config on users that are not
> interested in the python interface with boost?

I don't think there is any impact, each library is packaged into a separate
package named with boost-<lib name>

> What's the additional cost in compile time?

A quick test indicates that it's less than 1 second:

$ grep -r 'Elapsed time:' 20170823073306/boost-1.64.0-r0/do_compile 20170823075302/boost-1.64.0-r0/do_compile
20170823073306/boost-1.64.0-r0/do_compile:Elapsed time: 50.96 seconds
20170823075302/boost-1.64.0-r0/do_compile:Elapsed time: 51.63 seconds

Thanks,
Jackie

> 
> 
> 
> There certainly are a rush of package updates and other changes
> coming in; apologies for the lack of warning and late arrivals.
> Other than doing more testing and explaining what tests were done,
> what else can we do to help?
> 
> 
> ../Randy
> 
> 
> > Cheers,
> >
> > Richard
> >
> >
> 
> 
> --
> # Randy MacLeod. SMTS, Linux, Wind River
> Direct: 613.963.1350 | 350 Terry Fox Drive, Suite 200, Ottawa, ON,
> Canada, K2K 2W5