Latest ruby 1.8.5_1

[Paul Guyot]

Le 7 nov. 06 à 07:42, Jordan K. Hubbard a écrit :

>
> On Nov 6, 2006, at 2:02 PM, Paul Guyot wrote:
>
>> I don't know how to turn it into a variant in such a way that  
>> without this variant, ruby doesn't touch tk & tcl if they're  
>> available.
>
> Well, maybe if the trace code returned ENOENT on any attempt to  
> satisfy non-explicit dependencies, you could use it to create a  
> virtual chroot and then turn that virtual chroot mode on by  
> default.   Oh wait, we already went over all that in the message  
> you cited. :-) :-)

Heh. You love to be right, don't you?

I gave more thought to the way MacPorts work recently and I believe  
trace mode needs to be on by default. Then the problem is that it  
generates warnings where we want errors to make sure that portfiles  
are correct. In such a case, we want the minimum dependency set. For  
example, many ports will use MP install or autoconf where the system  
one would be perfectly fine. So yes, I changed my mind and I think we  
should have a chroot-like environment like you suggested -- it has  
some holes as ports could disable the dyld injection, but I guess  
it's fine for what we're doing, it's not a security concern, and it's  
much cheaper than a real chroot with union mounts, and it provides  
informations about forbidden accesses.

The problem is I don't have enough time to implement all this now. I  
toyed with ruby ports because of a work project of mine that is based  
on ruby. I'll try to do the 1.3.3 release as asked by James, but I  
think this will be all for 2006.

Paul
-- 
Ministre ultraplénipotentiaire en disponibilité.
Mobile. Sans baignoire fixe.
http://www.kallisys.com/
http://www-poleia.lip6.fr/~guyot/