Confirmed it's a difference in ido configuration - I'm using
ido-everywhere, not ido-ubiquitous.
On Thu, Jul 13, 2017 at 11:19 AM, Nathaniel Flath <flat0...@gmail.com> wrote:
> On Thu, Jul 13, 2017 at 11:02 AM, Kyle Meyer <k...@kyleam.com> wrote:
>> Nathaniel Flath <flat0...@gmail.com> writes:
>>
>> [...]
>>
>>>> I'm not an ido user, but testing with
>>>>
>>>> (require 'ido-ubiquitous)
>>>> (ido-ubiquitous-mode 1)
>>>>
>>>> partial completion seems to work fine when I hit tab after calling
>>>> org-agenda-filter-by-tag. Are you using ido-ubiquitous, or are you
>>>> using some other package to get ido completion for non-file/buffer
>>>> completion?
>>>
>>> The issue is that that 'tab' is required - in the previous version of
>>> org, this was not required and 'enter' would do the completion for
>>> you.
>>
>> I'm only using tab right after calling org-agenda-filter-by-tag to get
>> to the initial tag prompt, following the instructions in the minibuffer.
>> Isn't that always needed?
>>
>> Here is how I'm testing (with emacs -Q):
>>
>> * list TODOs with the agenda (org-agenda, follow by "t")
>> * on an entry, hit "/" for org-agenda-filter-by-tag
>> * hit tab to get to the tag prompt
>>
>> Once in the prompt, I can select a partial completion by just hitting
>> enter.
>
> I follow the same steps:
> * list TODOs with the agenda (org-agenda, follow by "t")
> * on an entry, hit "/" for org-agenda-filter-by-tag
> * hit tab to get to the tag prompt
> * type 'ho' + enter - this causes everything to disappear, because it
> is trying to filter for 'ho' instead of 'home' ('ho' is not a valid
> tag)
>
> I've upgraded to the lastest version of ido-ubiquitous and this
> behaviour persists.
> (This is also my understanding of what the expected behaviour should
> be, based on testing of completing-read and reading the function
> docs).
>
>>
>> I've executed the above steps with two setups:
>> release_8.3.6/org-completion-use-ido and release_9.0.9/ido-ubiquitous.
>> I couldn't spot any differences. Hitting enter for partial completion
>> worked in both cases.
>>
>>>> In this particular case, your change looks OK because I think all useful
>>>> values should be in the collection. But I don't understand why you're
>>>> not able to do partial completion without this change. Are you running
>>>> into the same issue everywhere else that completing-read is called with
>>>> a nil value for require-match?
>>>
>>> I haven't checked everywhere else - presumably.
>>
>> It'd be good to figure out why you're seeing different behavior than I
>> am. Perhaps the steps I list above aren't the right steps to trigger
>> it. Or maybe you have some non-default ido configuration. Or maybe the
>> difference comes from an ido version mismatch. (I'm using the ido from
>> Emacs 25.2 and ido-ubiquitous checked out at v3.14-4-ga1c2965.)
>
> I'm on emacs 24.5.1 - I'll try updating, but I suspect the problem will
> persist.
>>
>> As it stands, I can't reproduce the problem described in your commit
>> message.
>>
>> --
>> Kyle
