Could this not be solved with a new 'set' flag?
set -o tabtabcomp
This would always treat TAB-TAB as a completion request.
Just wonderin'...
Mario DeFazio
Satisfied ksh user since 1983 and holder of an American Bell medallion
;-)
Josh Hurst wrote:
> On 6/19/06, David Korn <dgk at research.att.com> wrote:
>
>>
>>
>> On 6/17/06, Martin Schaffstall <martin.schaffstall at googlemail.com> wrote:
>> > On 6/16/06, Glenn Fowler <gsf at research.att.com> wrote:
>> > >
>> > > changes and notes:
>> > > (2) tab-tab as requested plus tab-tab-#-tab to disambiguate from list
>> >
>> > Yes, this works now very well - with the sole exception that tab-tab
>> > in an empty area still inserts two tab characters instead of a list of
>> > choices in the current directory :-(
>>
>>
>> This behavior is intensional. Many current users expect a TAB to mean
>> TAB
>> especially when entering compound commands from the terminal.
>> I suspect that2 is is less of an inconvenience for a user to type one
>> character before doing a completion, then it is for a user to
>> hit ^V before each tabr especially with multiple tabs.
>
> But how many users really need that? Other shells such as bash can
> happily live with ^V+TAB and exclusively uses TAB and TABTAB for
> command and variable completion. And users like that implementation -
> the last time it broke in Debian caused a revolt against the package
> maintainer.
>
> Maybe we make a compromise: -o emacs retains the current behaviour
> while -o gmacs treats TAB and TABTAB like Martins patch. Will that
> work for you?
