Bug#411635: getopt_long_only: short vs. long option ambiguity

[Alexander Gattin <xrgtn@yandex.ru>]

Hello,

I disagree on 2 points:

1. The bug should be assigned to libc6 instead of libc6-i686,
   because it's equally applicable to all architectures,
   including i386 of course.

2. The severity isn't minor in any way, because:

   * the bug potentially affects a lot of packages that
     use getopt_long_only()

   * it introduces a whole new class of ambiguities that
     were not even mentioned in getopt's documentation _and_
     were ignored in implementation altogether, AFAIS

   * the current getopt_long_only()'s behaviour is completely
     against any expectation or common sense, that greatly
     increases risk of programming errors and bugs

Citing man getopt_long_only:
>        getopt_long()  and getopt_long_only() also return the
>        option character when a short option  is  recognized.
>        For  a  long option, they return val if flag is NULL,
>        and 0 otherwise.  Error and -1 returns are  the  same
>        as  for  getopt(), plus ’?’ for an ambiguous match or
>        an extraneous parameter.

   * there's no "option `-l' is ambiguous" warning message
     generated at all, or '?' returned for short vs. long
     option ambiguity. (you get the '?' and warning when
     there's amgiguity between 2 long options (e.g. '-longa'
     vs. '-longb')).

   * curent resolution of ambiguities is inconsistent, because
     it prefers long option in one case and short one in all
     others (see below):

There's no description about how the current implementation
resolves short vs. long option ambiguities, and the actual
rule is the next (JFYI):

a) lone (unbundled) short option
   '-l':  '-l'  vs.  '--long'

   => short option takes precedence

b) bundled short option in beginning of a bundle
   '-ls':  '-l' '-s'  vs.  '--long' '-s'

   => short option is chosen ('-l' '-s')

c) bundled short option in middle of a bundle
   '-Sls':  '-S' '-l' '-s'  vs.  '-S' '--long' '-s'

   => short option is chosen ('-S' '-l' '-s')

d) bundled short option at end of a bundle
   '-sl': '-s' '-l'  vs. '-s' '--long'

   => long option is preferred here ('-s' '--long')

-- 
IMHO this doesn't qualify as a minor bug.
Regards,
xrgtn