[Bash-completion-devel] [bash-completion-Bugs][311384] DBTS 336987: shell variables should always be a valid completion when syntactically valid

[bash-completion-bugs at alioth.debian.org]

Bugs item #311384, was opened at 30/01/2009 09:21
Status: Open
Priority: 3
Submitted By: David Paleino (hanska-guest)
Assigned to: Nobody (None)
Summary: DBTS 336987: shell variables should always be a valid completion when syntactically valid 
Distribution: None
Originally reported in: Debian BTS
Original bug number: 336987


Initial Comment:
Normally, bash can complete shell variables when appropriate; for
example, "foo=1" followed by "echo $f<TAB>" will complete to "echo
$foo".  However, programmable completion interferes with this; for
example, "svn ls $f<TAB>", "cvs $f<TAB>", and "ssh $f<TAB>" all fail to
complete $foo.  Shell variables should always be a valid completion when
they are syntactically valid, since they could always contain a string
appropriate for the current command.

(By "when they are syntactically valid", I mean that the shell already
knows not to complete "echo \$f<TAB>" into "echo \$foo", and that
shouldn't change.)

----------------------------------------------------------------------

You can respond by visiting: 
http://alioth.debian.org/tracker/?func=detail&atid=413095&aid=311384&group_id=100114