Keyman Glide (Swipe) Typing

Dear Keyman Developers,

Please add Glide typing (Swipe typing) method possibility.

Thanks in advance.

Welcome to the community, @Vasyl.

The feature is not currently on our roadmap, but I’ve created an issue to track it at feat(android/iOS): Keyman Glide (Swipe) Typing - OBU · Issue #8233 · keymanapp/keyman · GitHub.

1 Like

There are a number of reasons that we have not yet implemented this:

  1. Our current gesture support is fairly limited. We’re currently placing higher priority on multitap input and flicks, as they’re simpler gestures.
  2. A good swipe-input implementation benefits greatly from a good predictive-text implementation.
    • One issue here… is that we don’t have very many user-submitted dictionaries yet. Without that, swipe-input needs much higher precision from users to work properly than might be expected.
    • We would likely not enable swipe input for any language without a matching dictionary.
1 Like

Yes, of course. Effectiveness of swipe typing wholely depends on predictivenes of dictionary. And should be developed in parallel.

Everyone can self-check this by comparing modern swipe keyboards (Gboard, SwiftKey) with old swipe keyboards (SlideIT, Swype+Dragon). Weak dictionary = weak swipe typing.

For the last 10 years everyone is using only swipe typing which saves time and eliminates misprints.

Without swipe typing Keyman will be interesting only for the narrow circle of linguists and for the oldest generations.

As per my opinion, Swipe typing is the first priority to be implemented in Keyman if we want Keyman to be practically used in real everyday life by millions of people.

Swipe typing only works with a very high quality dictionary. If you can solve that for the thousands of languages we work with, then great, we can implement swipe in Keyman.

1 Like

Dear Marc,

I think that nobody deadly needs dictionaries for all thousands of languages. This is already done by Google and Microsoft and available in their universal keyboards. Most probably their big dictionaries were created by parsing the Wikipedia offline archive databases in those languages.

By the way Keyman Dev Team can do the same, if they think that they really need thousands of own dictionaries.

As per my option currently the exclusive niche of Keyman is to help minor nations, the smaller group of users with only one their own seldom language, which is not yet covered by universal keyboards of Google and Microsoft. And it’s those users’ responsibility to make dictionary in that seldom language. Keyman Dev Team should not try to build seldom language dictionary. Users will have to build dictionary by themselves. While Keyman should allow swipe and be practically usable modern keyboard. First Keyman Dev Team provides good wheels and then users will go ahead by themselves.

Anyway, your existing English dictionary for 200k lines is already enough predictable? Why not to see how swipe will work with it ? After that we will know how much lines dictionary is enough for swipe.

Unfortunately even our English dictionary is really not up to the task. It needs a significant investment to improve the quality and word selection.

1 Like

Dear Marc

Let’s say if I have a language dictionary with total 2.8 millions words of which however only 500k are counted, the rest are not. Will such big dictionary allow itself to use effective swipe or there are existing other limiting reasons ?

A large dictionary is not sufficient. You also need a grammar or phrase-level predictions, otherwise the system will always be predicting just the most common words, which will make it very hard to swipe coherent sentences reliably.

1 Like

About that. Keyman’s primary niche is to serve and facilitate typing in languages that Wikipedia and the major computer vendors don’t even serve yet. There would likely be no articles to parse for hundreds, if not 1000+, of the languages we currently target. Those are the “millions of people” we care about most - those language communities who have nobody else supporting them at all. That’s why we’re open-source.

Also, keep in mind that as an open-source project… we don’t exactly have server farms with tons of spare computing power to throw at the problem, even if such articles did exist.

I realize the quoted block was part of an exaggerated statement and thus is hyperbole, but TIL that at the ripe old age of 37, I’m part of “the oldest generations.” Granted, yes, I shouldn’t let my personal biases affect things.

Allow me to revisit one of my earlier points, though:

Our current gesture support is fairly limited. We’re currently placing higher priority on multitap input and flicks, as they’re simpler gestures.

Swipes are far more complex than the gestures listed in that quote. We need to make sure we get the simple things working correctly before jumping straight to the complex stuff. That said, while not yet fully implemented, we have made sure that the design we’re using for the new gestures will be compatible with possible future implementation of swipes.

1 Like

Dear Joshua

Sorry, I didn’t mean what you assumed. Even my age is more than yours. I meant the oldest people who like push-button retro phones without touchscreens. And if phone has touchscreen then keyboard should behave like on old phones without swipe. Thus old people will thank for the absence of swipe.

Anyway, Keyman is a great app, thanks to you and all Dev Team.

This thread is now closed. For future update on the feature request posted on GitHub, let’s track it there.