Ulrich Mueller
2018-03-09 10:30:42 UTC
The current EAPI 7 draft introduces rmread, rmwrite, rmpredict, and
rmdeny in addition to their add* counterparts.
This looks like a short-sighted solution and I believe that it is
going into the wrong direction. It is entangled with the details of
the current sandbox implementation. So if new features like network
sandboxing should be added later, we'll either need another set of
commands for those, or we'll end up with removing the rm* commands
again, in favour of a more consistent interface.
Since the rm* commands aren't urgently needed (the only things that
could profit from them are the live eclasses which are a corner case
by themselves), I suggest to omit this feature from EAPI 7 and rethink
the whole sandbox API for EAPI 8.
Ulrich
rmdeny in addition to their add* counterparts.
This looks like a short-sighted solution and I believe that it is
going into the wrong direction. It is entangled with the details of
the current sandbox implementation. So if new features like network
sandboxing should be added later, we'll either need another set of
commands for those, or we'll end up with removing the rm* commands
again, in favour of a more consistent interface.
Since the rm* commands aren't urgently needed (the only things that
could profit from them are the live eclasses which are a corner case
by themselves), I suggest to omit this feature from EAPI 7 and rethink
the whole sandbox API for EAPI 8.
Ulrich