Jolla c2 - predictive text input - howto?

Feature Requests
#1

With jolla c2, we no longer have predictive text input. I hoped things would be quickly fixed, but it looks like it’s taking time. And it has proven to be a lot more annoying than what i supposed it would be.

So, the alternative seems to be installing presage, from chum (any other alternative, not involving copying binaries from an old phone, welcomed).

Well, i tried to install presage from chum gui, and nothing changed, I could not get it to do anything, in fact, no app showed up and no new keyboard in settings. Are there some detailed instructions somewhere on how to setup presage for the C2 ?

2 Likes
#2

Not really an answer but for (more) information here you can found the discussion of last november.

4 Likes
#3

Thanks, it indeed helped. I now have a working english layout, now trying to create the database for french.

Will post detailed instruction and PR when it’s ready (hopefully before end of month…).

4 Likes
#4

More on this, it seems like text2ngram has terrible utf-8 support.

Before going on a large refactor which is needed to support utf-8, how did the others do to generate the databases ? Convert the corpus to a local encoding, generate the database, and then convert it back to utf-8 ? Or do the currently supported languages not use special characters ? (i doubt it for german).

#5

Strange, when i compiled with txt2ngram databases of my language - there was no problem with encoding, i even didn’t think that can be problem here) Did it on Ubuntu 22.04.

And by the way… Even with huge databases its almost impossible to type normally with Presage keyboard. Very poor prediction, very bugged correcting wrong letters. Better without it. Tested for long time. Xt9 by Jolla is much much better. Not excellent, but can be used for everyday typing.

#6

22.04 makes sense, because cf67f3b9, which breaks parsing anything that is utf8 source, is from november 2022. I could revert it, but the resulting database will be of very poor quality given my corpus (lots of multibytes word separators).