Re: --build-as-nobody by default.

Am 14.04.2012 um 17:20 schrieb Alexander Hansen <[email protected]>:

> On 4/14/12 6:58 AM, Dustin Cartwright wrote:
>> 
>> On Fri, Apr 13, 2012 at 5:52 PM, Max Horn <[email protected]> wrote:
>> How about this strategy instead: First we turn off BuildAsNobody globally again by default. Then we make a release with the new BuildAsNobody field and command line option in it. *Then* we email to -devel and ask all authors to check their packages (and how) and add BuildAsNobody as needed. Meanwhile, we update all base packages as necessary. Finally, after a certain grace period (I'd suggest a month or two at least), we can turn off BuildAsNobody globally. Some packages will still break, but at least this way we give maintainers a fair chance to fix their act. As opposed to forcing them to fix their packages IMMEDIATELY after the release or else potentially getting tons of user complaints for something that might not even be avoidable from their side. In addition, this approach might reduce the pain for our users a bit, too... ;).
>> 
>> Hello,
>> 
>> I agree that it would be a good idea to have a transition period during which package maintainers can add "BuildAsNobody: false" as needed before releasing a version of fink which builds as nobody by default. As you say, this would mean modifying the validator and notifying fink-devel well in advance of the change.
>> 
>> However, I don't see the point of going through a stage where BuildAsNobody is off by default, but opt-in at the package level.

Neither do I :). Which is why I never suggested this (just saying, in case people misinterpreted me).


>> It wouldn't really lead to more packages getting tested because maintainers already have the ability to test their packages with the --build-as-nobody flag.

Exactly.

>> The poorly maintained packages still won't get tested until the default is switched. That's why it seems like a good idea to have BuildAsNobody the default in master: it makes it more likely that problematic packages will be first noticed by someone who recognizes the problem rather than a random user.

Actually, it wouldn't, as hardly anybody runs a system using our master branch. The wide testing would only come with a full release. But by that time, this woud affect all users, so by then we should have prepared the system already. This is one of the drawbacks of having abandoned the unstable tree. (of course the advantages of abandoning unstable outweight this by far).

>> 
>> Dustin
>> 
>> 
> I'd argue that we've been in the transition period since --build-as-nobody was established. :-) 

yes and no...

> We've been identifying packages that don't build as nobody and fixing those that can all along, but haven't had a .info file field name to mark them.

Because while people could test their packages with b-a-n, those who had packages which really needed to be built as root had no way to do anything about it. Expecting them to now do this instantly seems wrong.

> 
> Since fink ignores .info file fields it doesn't recognize (at least for booleans), maintainers can mark their appropriate packages with "BuildAsNobody: false" right now.

but then they fail validation. I still think that's the wrong way, and the only "advantage" it has is that it allows us is that we give in to impatience.

Really, the proper way is so simple: disable ban by default, make a release, then give maintainers a last stern warning and a grace period. 

Cheers,
Max
> -- 
> Alexander Hansen, Ph.D.
> Fink User Liaison
> http://finkakh.wordpress.com/2012/02/21/got-job/
> ------------------------------------------------------------------------------
> For Developers, A Lot Can Happen In A Second.
> Boundary is the first to Know...and Tell You.
> Monitor Your Applications in Ultra-Fine Resolution. Try it FREE!
> http://p.sf.net/sfu/Boundary-d2dvs2
> _______________________________________________
> fink-core mailing list
> [email protected]
> List archive:
> http://news.gmane.org/gmane.os.apple.fink.core
> Subscription management:
> https://lists.sourceforge.net/lists/listinfo/fink-core

------------------------------------------------------------------------------
For Developers, A Lot Can Happen In A Second.
Boundary is the first to Know...and Tell You.
Monitor Your Applications in Ultra-Fine Resolution. Try it FREE!
http://p.sf.net/sfu/Boundary-d2dvs2

_______________________________________________
fink-core mailing list
[email protected]
List archive:
http://news.gmane.org/gmane.os.apple.fink.core
Subscription management:
https://lists.sourceforge.net/lists/listinfo/fink-core