I'd like to be a tester of busybox

[Paul Fox]

denis wrote:
 > I am curretly do not even perform cross-compiles, so I feel this
 > area is sort of not well covered.

happily, many others of us do this all the time.  cross-compiling
is pretty well covered, i suspect.

 > 
 > If you are desktop guy - try to replace as much of desktop tools
 > with relevant busybox applets as practical. Expect some breakage.
 > For example: ls -i is not supported.

i really can't agree with this advice.  not if you're a serious
developer, and your desktop is important to you.  the utilities
in busybox just aren't ready for prime time, in my opinion.  (and
i'm speaking as someone who has shipped a bunch of product that
uses it, has commit access, and who regularly contributes
patches.)  error checking is missing, command-line options are
missing, features are unimplemented, or mis-implemented, etc.

by all means, do busybox testing with the goal of turning busybox
into a desktop replacement, but if you're just trying to "help
out", that's a bigger risk than i'd be willing to take (denis'
apparent success notwithstanding).

here's a minor example that we noticed today: "busybox tar -tv"
doesn't show the major/minor numbers of device nodes from the
archive.  this isn't something that would probably be a problem
on an embedded system (and if it did, it would hopefully be
caught in testing) but on a desktop, it could cause some user
confusion or frustration.  there are lots more like this.

i'm _not_ saying busybox is no good.  i'm saying that one should
be well aware of it's (many) limitations before putting it to a
task that it may not be ready for.

 > 
 > The more things you put to use, the more bugs you expose :)

now _that_ i can agree with.  :-)

paul
=---------------------
 paul fox, pgf at brightstareng.com