Skip site navigation (1)Skip section navigation (2)
Date:      Mon, 09 Jun 2014 18:00:15 +0200
From:      Guido Falsi <mad@madpilot.net>
To:        sbruno@freebsd.org, Alex Dupre <ale@FreeBSD.org>
Cc:        ports list <freebsd-ports@freebsd.org>
Subject:   Re: gcc vs gcc48
Message-ID:  <5395DA0F.1000107@madpilot.net>
In-Reply-To: <1402328895.76875.11.camel@bruno>
References:  <1402327321.76875.8.camel@bruno> <5395D387.8020902@FreeBSD.org> <1402328895.76875.11.camel@bruno>

next in thread | previous in thread | raw e-mail | index | archive | help
On 06/09/14 17:48, Sean Bruno wrote:
> On Mon, 2014-06-09 at 17:32 +0200, Alex Dupre wrote:
>> Sean Bruno ha scritto:
>>> Ended up in a weird position today running a build.  My poudriere
>>> instance ended up building gcc and gcc48 at the same time.
>>>
>>> I thought this was odd.  Shouldn't gcc48 supplant gcc for requirements?
>>
>> Every port is built independently from the others, so gcc48 is not found
>> when building other ports that doesn't strictly require it.
>>
> 
> Understood.
> 
> lang/gcc isn't successfully building at the moment.  So, when I restart
> this build, gcc48 will be detected as built so it will be pulled in.
> Which will in turn start building the ports that were failed on the
> previous run.

No, not in poudriere, in poudriere it will simply notice the gcc48
package is already build and instal that in the jail for the ports
needing it. For ports which were calling lang/gcc it will still try to
build lang/gcc, since it failed and there is no ready to install package.

Was your description what you actually saw happening? this would be very
strange.

> 
> This seems like a broken "feature" of ports IMO.  I.e. non-consistent
> runs.

It is a bad feature of ports on a live system, poudriere avoids
this...at least it should.

-- 
Guido Falsi <mad@madpilot.net>



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?5395DA0F.1000107>