Skip site navigation (1)Skip section navigation (2)
Date:      Mon, 27 Feb 2012 14:52:31 -0800
From:      Doug Barton <dougb@FreeBSD.org>
To:        Olivier Smedts <olivier@gid0.org>
Cc:        freebsd-ports@freebsd.org
Subject:   Re: portupgrade -> portmaster Rosetta Stone?
Message-ID:  <4F4C092F.8080804@FreeBSD.org>
In-Reply-To: <CABzXLYN%2Byb_dZcYPJegXD9_tk49KjdRcH1oj1QJgMXBfnmdGyA@mail.gmail.com>
References:  <CA%2BE3k93doEmK7YpiZLgBCB5WUmOeCCLLLQBak_-Nx8sGqiafhQ@mail.gmail.com> <CABzXLYM47FjwmEvKijOp41Hn0KhTTPJ-dnX58PxTS8oUff-THQ@mail.gmail.com> <4F4BA7CE.20107@FreeBSD.org> <CABzXLYMVf3JU6Q=mXHV3ddPvKjWmNtWKsbcTWy9FjPOnNYoMLw@mail.gmail.com> <CAF6rxgnNHBOt8NibEr=Vgmqw9bTPmG7tG7s6e-dAHrtbbtjXOA@mail.gmail.com> <4F4BCF87.6070209@FreeBSD.org> <CABzXLYP==pMzme6yD-g8wosff3YHDR_%2BSeaBQuQtaHyocJoFPA@mail.gmail.com> <4F4BE02D.4020907@FreeBSD.org> <CABzXLYN%2Byb_dZcYPJegXD9_tk49KjdRcH1oj1QJgMXBfnmdGyA@mail.gmail.com>

next in thread | previous in thread | raw e-mail | index | archive | help
On 2/27/2012 12:04 PM, Olivier Smedts wrote:
> 2012/2/27 Doug Barton <dougb@freebsd.org>:
>> On 2/27/2012 11:53 AM, Olivier Smedts wrote:
>>> What about a command line flag to ignore errors during backup package
>>> creation?
>>
>> What about ... no. :)
> 
> Don't get me wrong, portmaster is a great tool, reliable, and I'm
> using it nearly daily.
> 
> But why not, if there's a reason ? "potential foot shooting" ?

It's a general philosophy thing that is common to mergemaster and
portmaster. I feel *incredibly* strongly that it's important for key
system management tools to *not* make unsafe assumptions. I (and by
extension the tools that I write) have no way to know what changes are
mission-critical to a given user. Therefore, I have *no business*
merrily blowing stuff away that the user may have been depending on.

In the particular case of creating backup packages, with or without the
-b option, the default assumption is that those packages are being
*relied* on by the user to roll back an update in case of catastrophe.
For users that don't care about what happens with the backup packages
there is already a command line option to disable creation of them. My
assumption is that if the user is not using the option to disable them
altogether that the backup packages must have value. (Also, a
meta-issue, if the ports developers are doing their jobs then backup
package creation failure should be an incredibly rare occurrence.)

OTOH, I got many requests from users to add a knob to bypass the warning
about failed package creation. Personally, I don't like this knob, and
would never use it. But in order to please the user base I added the
feature, but included it as an "expert" option that requires
configuration in an rc file.

Frequently the next response to this line of reasoning is to insist that
we're all grownups, and foot-shooting options should not be disallowed
by policy. Well, fine. :)  Go enable that option and blow off all the
toes you want. My point is that the kinds of people who want a knob like
that (expert users, frequent updaters, etc., and, not to hurt your
feelings, but you're a very small percentage of the userbase) are
precisely the same target audience that ought to be able to read the man
page and configure expert options.


hth,

Doug (bet you wondered how I was gonna work that in, didncha)

-- 

	It's always a long day; 86400 doesn't fit into a short.

	Breadth of IT experience, and depth of knowledge in the DNS.
	Yours for the right price.  :)  http://SupersetSolutions.com/




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?4F4C092F.8080804>