[Feature Request] Split learning words per languages.

[Mlocik97]

Hello,

today when I use typing booster and set lang to en, I write some english words, but then after switching to lang to sk, it suggest me words learned in English. Can this be split?

Thanks.

[mike-fabian]

@Mlocik97

Since I fixed this issue https://github.com/mike-fabian/ibus-typing-booster/issues/570

with the release of https://github.com/mike-fabian/ibus-typing-booster/releases/tag/2.27.0 it is possible to configure two typing-booster engines for different languages.

For ibus-typing-booster >= 2.27.0 there is still the same typing-booster engine as before:

$ ibus list-engine | grep booster
  typing-booster - Typing Booster

but in addition to that there are not many tb:language:name engines setup by default to emulate ibus-m17n for a specific m17n-db input method. I.e. only do transliteration, don't complete anything, don't record anything, don't show any candidates.

But these are only the default settings! Any of these engines can be configured freely to reenable all the features of the full typing-booster.

So it is now possible for example to use the main typing-booster engine for English and choose any other engine for example this one:

$ ibus list-engine | grep tb:sk 
  tb:sk:kbd - sk-kbd (Typing Booster)

and configure that for another language by adding any input methods and dictionaries you like and

• Uncheck the option ☐ Enable suggestions by key
• Uncheck the option ☐ Off the record mode
• Bind the key Tab again to the command enable_lookup
• Remove the keybinding Left from the command commit_and_forward_key
• Set Preedit underline to single or whatever style you prefer

See also the chapter:

Simulate the behaviour of ibus-m17n https://mike-fabian.github.io/ibus-typing-booster/docs/user/#2_3

I.e. you can chose any of the engines simulating ibus-m17n and then change the settings to the “normal” typing-booster default settings and then you have yet another “normal” typing-booster engine which you can use for another language.

This other engine stores the learned data in a different user database depending on the name of that engine. For example at the moment I have a few user dababases for different typing-booster variants:

$ ls ~/.local/share/ibus-typing-booster/user*db
/home/mfabian/.local/share/ibus-typing-booster/user-hi-inscript2.db
/home/mfabian/.local/share/ibus-typing-booster/user-hi-itrans.db
/home/mfabian/.local/share/ibus-typing-booster/user-t-latn-post.db
/home/mfabian/.local/share/ibus-typing-booster/user-t-rfc13445.db
/home/mfabian/.local/share/ibus-typing-booster/user-t-rfc1345.db
/home/mfabian/.local/share/ibus-typing-booster/user.db

user.db is for the typing-booster engine, user-hi-itrans.db is for the tb:hi:itrans engine etc....

That means learning of words for different languages can be stored in different databases now.

But on has to switch engines when switching languages.

Using only one typing-booster for several languages may sometimes give suggestions from the wrong langauge. Although if often gets it right because it looks at the context of the last words typed, if the last 2 words are from langauge A it is more likely that one will get suggestions from langauge A.

So in most cases I think using only one engine saves more keystrokes by avoiding to switch engines, even though one occasionally may get suggestions from a wrong language.

You can try whether switching engines for different langauges is more efficient for you, I would like to hear how this works for you.

[Mlocik97]

Thank you, will check and try it.

[mike-fabian]

@Mlocik97

Since I fixed this issue #570

with the release of https://github.com/mike-fabian/ibus-typing-booster/releases/tag/2.27.0 it is possible to configure two typing-booster engines for different languages.

For ibus-typing-booster >= 2.27.0 there is still the same typing-booster engine as before:

$ ibus list-engine | grep booster
  typing-booster - Typing Booster

but in addition to that there are not many tb:language:name engines setup by default to emulate ibus-m17n for a specific m17n-db input method. I.e. only do transliteration, don't complete anything, don't record anything, don't show any candidates.

But these are only the default settings! Any of these engines can be configured freely to reenable all the features of the full typing-booster.

So it is now possible for example to use the main typing-booster engine for English and choose any other engine for example this one:

$ ibus list-engine | grep tb:sk 
  tb:sk:kbd - sk-kbd (Typing Booster)

and configure that for another language by adding any input methods and dictionaries you like and

* Uncheck the option `☐ Enable suggestions by key`

* Uncheck the option `☐ Off the record mode`

* Bind the key `Tab` again to the command `enable_lookup`

* Remove the keybinding `Left` from the command `commit_and_forward_key`

* Set `Preedit underline` to `single` or whatever style you prefer

See also the chapter:

Simulate the behaviour of ibus-m17n https://mike-fabian.github.io/ibus-typing-booster/docs/user/#2_3

I have changed the options a bit in the 2.27.27 release. The online documentation in

https://mike-fabian.github.io/ibus-typing-booster/docs/user/#2_3

is already updated.

The “restricted” engines (I.e. those starting with tb: when listed by ibus list-engine) now use ☑️ Enable suggestions by key by default and also use the default for the command enable_lookup (which is ['Tab', 'ISO_Left_Tab']).

• Check the option ☑️ Word predictions (assuming you want word predictions)
• Check the option ☑️ Unicode symbols and emoji predictions (optional, only if you like)
• Add some emoji trigger character (optional, only if you like this for searching emoji)
• Uncheck the option ☐ Off the record mode
• Remove the keybinding Left from the command commit_and_forward_key
• Set Preedit underline to single or whatever style you prefer

I think it is more logical now.