Skip site navigation (1)Skip section navigation (2)
Date:      Fri, 28 Feb 2014 09:29:05 +1100
From:      Dewayne Geraghty <dewayne.geraghty@heuristicsystems.com.au>
To:        marino@freebsd.org, Baptiste Daroussin <bapt@freebsd.org>
Cc:        "freebsd-ports@freebsd.org" <freebsd-ports@freebsd.org>
Subject:   Re: Support for pkg_*
Message-ID:  <530FBC31.1050408@heuristicsystems.com.au>
In-Reply-To: <530D18DE.6020003@marino.st>
References:  <530C5793.2070208@heuristicsystems.com.au> <20140225141144.GA87810@spectrum.skysmurf.nl> <20140225163457.GI83610@ithaqua.etoilebsd.net> <20140225220844.GA92169@spectrum.skysmurf.nl> <530D18DE.6020003@marino.st>

next in thread | previous in thread | raw e-mail | index | archive | help

On 26/02/2014 9:27 AM, John Marino wrote:
> On 2/25/2014 23:08, A.J. 'Fonz' van Werven wrote:
>> Baptiste Daroussin wrote:
>>
>>> Can we stop advertising the above, this is completly wrong, it hides the
>>> dust behind the carpet and won't fix anything!
>>>
>>> The said port is needed a fix.
>> Granted: it's not staging itself that is the problem, it's incorrect usage
>> thereof. But anyone possessing even the tiniest trace amount of realism
>> will have to concede that port maintainers do occasionally get it wrong.
>> And when they do the errors as reported by the OP are a tell-tale symptom.
> No one should have a problem conceding that.  The issue is that it seems
> that many ports are staged without checking in redports / poudriere or
> otherwise.
>
> I was even guilty of this the other day.   I was on the road and I
> created two new ports that easily passed with DEVELOPER_MODE.  Neither
> built in clean environment though and I got rewarded with pkg-fallout
> messages.
>
> A lot of stuff gets committed that it's clear was never remotely tested,
> not even with portlint.  But don't blame the tools -- the port needs to
> be fixed.  I agree that we should never advise "NO_STAGE=yes", ever.  If
> the port is broken, so be it.  PR, patch, normal process.
>
> There's been a lot of understandable grumbling due to growing pains of
> major infrastructure changes by users, so telling users to revert these
> changes isn't a good look.  Let's just try to get the port fixed in a
> reasonable timeframe (e.g. get the guy that broke it to take care of it).
>
> John
>
Good point John, but its a seeping breakage - and most likely an
inadvertent constraint from /usr/ports/Mk.  I PR'ed net/rsync in Nov
2013, openssl and monit in February (both added staging); and I'm
confident that if I extended the "base" system with other ports rooted
in /usr that there would be others.

 Its always a tar stat related failure and the ONLY way to get ports,
that use PREFIX=/usr to build their package, is to also add MANPREFIX=/usr.

Ref:
openssl, monit - http://www.freebsd.org/cgi/query-pr.cgi?pr=187076
rsync - http://www.freebsd.org/cgi/query-pr.cgi?pr=183669

Regards, Dewayne.



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?530FBC31.1050408>