On 08/27/2013 05:40 PM, Andreas Mohr wrote: > after some progress with debirf (i.e., a rootfs finally successfully created), > I decided to start using debirf -s (WRITE_MODE=skip) > yet by simply having it *inserted* to my prior cmdline without realizing > that -s WRITE_MODE setting would get overwritten by my leftover > subsequent -n (WRITE_MODE=rewrite) or -o (WRITE_MODE=overwrite) options. > --> Result: debootstrap called again despite not wanting it this time... > > while true ; do > case "$1" in > ... > -n|--new) > WRITE_MODE=rewrite > shift 1 > ;; > -o|--overwrite) > WRITE_MODE=overwrite > shift 1 > ;; > -s|--skip) > WRITE_MODE=skip > shift 1 > ;; > ... > > (witness --skip getting overwritten by any subsequent --new / --overwrite > in this while loop) > > > The root cause of my confusion case is that -n -o -s parameters are > wholly redundant: instead, they are supposed to be a *single* *mode* switch, > to cleanly reflect exactly how it's implemented within-script > (WRITE_MODE=xxx), > since this seems appropriate here. > > These redundant options should thus be deprecated in favour of e.g. a combined > -m (--mode) switch taking rewrite/overwrite/skip arguments, I'd think > (and probably add a deprecation warning for a while, > whenever the old switches happen to get used).
If i understand correctly, checking for redundant options and
terminating with a non-zero error code and a warning would have avoided
your problem without breaking compatibility with existing scripts that
use only one of the options.
I'd welcome a patch that implements such a check. I'd rather not add a
new command-line option (e.g. --mode) if we can avoid it.
Thanks for pointing this out!
--dkg
signature.asc
Description: OpenPGP digital signature
