Noam Postavsky <npostavs@users.sourceforge.net> writes:

> On Mon, Nov 28, 2016 at 3:59 PM, Mark Oteiza <mvoteiza@udel.edu> wrote:
>>>
>>> npostavs@users.sourceforge.net writes:
>>> >
>>> > I tentatively suggest the patch below, but I want to add some tests
>>> > before commiting anything.
>>>
>>> While adding tests, I found another inconsistency: when given a
>>> hashtable with symbol keys, test-completion passes the symbol-name to
>>> PREDICATE, while all-completions and try-completion pass the original
>>> symbol key.  Here are two patches, the first for this bug, and the
>>> second for the other inconsistency.
>>
>> The first hunk of 2/2 isn't applying here.
>
> Oh, that's probably because I generated it with -w, so some whitespace
> changes are missing in 1/2. I'll post a full patch tonight.

Here are the same patches with full whitespace changes.