[Exherbo-dev] Do we really need addpredict?

Ingmar Vanhassel ingmar at exherbo.org
Fri Aug 21 22:01:53 BST 2009


Excerpts from Ali Polatel's message of Fri Aug 21 22:47:29 +0200 2009:
> $subject says it all.
> Do we really need addpredict? If so why?
> Currently there are two places where it's used.

On top of that there're 5 packages in official repositories that still
use addpredict.  Those would need to be fixed.

> The first place is the default_src_test which adds "/" to
> predict prefixes. The reasons for this is historical. Removing it
> doesn't seem to make more tests fail. Although I must admit I haven't
> tried to build all tree with this disabled.
> 
> The second place is scm-cvs exlib. Here it's only used to suppress an
> access violation but sydbox master has support for access violation
> hooks which are just used for this purpose. Here's how you can use them
> to suppress an access violation:
> 
> cat >syd.hook <<EOF # Such hooks will usually end up in $FILES
> echo "Pid: $1"
> echo "System call name: $2"
> echo "Path that caused the access violation: $3"
> if [[ "$2" == open && "$3" == "/etc/passwd" ]]; then
>         exit 1 # Hook returning non-zero prevents the access violation
>                # from being raised. The access is _still_ denied though.
> fi
> EOF
> chmod +x syd.hook
> sydboxcmd addhook "$(pwd)"/syd.hook # I'll probably add an addhook
>                                     # command, that's aliased to this.
> [run the stuff that raises the useless access violation here]
> sydboxcmd rmhook "$(pwd)"/syd.hook  # And an rmhook command as well.
> 
> So we have no use cases left for addpredict.
> Please start whining or I'll remove addpredict soonish.

No objections, once packages using it are fixed not to need it anymore.

-- 
Exherbo KDE, X.org maintainer



More information about the Exherbo-dev mailing list