Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Kanjipad is too strict on errors #787

Closed
Akinori0713 opened this Issue Aug 9, 2017 · 15 comments

Comments

Projects
None yet
2 participants
@Akinori0713
Copy link

Akinori0713 commented Aug 9, 2017

Thank you for making the time to report a bug and make Aedict better!

Thank you for your interest in Aedict. Prior submitting bugs, please make sure that you have newest Aedict from Google Play, Blackberry store or Amazon App store.

A screenshot is often worth a thousand words, so please make sure to include one ;)

Income from Google Play and other stores is rather small, and is made even smaller by the 30% fee charged by Google and 20% VAT charged in most countries. Please consider supporting the project on Patreon:

https://www.patreon.com/bePatron?u=2914375

@mvysny

This comment has been minimized.

Copy link
Owner

mvysny commented Aug 9, 2017

That is true - internal Kanjipad uses a rather very simple algorithm which is quite unforgiving on stroke order, and, most importantly, stroke count. There is a possibility to use the KanjiPad Extension which should be way more forgiving: http://aedict.eu/kanjipadextension/index.html

@mvysny mvysny self-assigned this Aug 9, 2017

@Akinori0713

This comment has been minimized.

Copy link
Author

Akinori0713 commented Aug 9, 2017

@mvysny

This comment has been minimized.

Copy link
Owner

mvysny commented Aug 9, 2017

It is not as easy to use as a few other apps on iOS like "Shirabe Jisho".
It'd take me minutes just to redraw the kanji again and again in different
stroke orders to get the kanji I want and sometimes I'd just give up and
type in the romanized form of the word itself for the kanji.

I'm sorry to hear that. Is it possible to learn which algorithm does Shirabe Jisho use? Perhaps it uses some kind of open/free algorithm which could then be integrated into Aedict...

That said, I really love the app. It is honestly the best Japanese
Dictionary on Android I've ever used while I do understand it still needs
improvements.

Thanks, and it is indeed true that there's a room for improvement.

the GUI looks a bit quite old with all the Material Design stuff going nowaday you know

True. Unfortunately I am no UX designer and I have no money to hire UX designer... I'd love to have a newest material design, but I need concrete hints as on which part of the app to improve. The problem with Material Design is that:

  • it contains questionable UX guidelines (such as buttons at the top, basically unreachable with your thumb)
  • it wastes a lot of space, which is in conflict with Aedict philosophy of cramming lots of information onto one screen

So, it's not easy to do the design right; the low income from Aedict forbids me to hire an UX guy. From 10 eur I charge, the state will take 20% VAT and Google will take 30% as a fee; Aedict is selling really slowly nowadays so I earn like ~80 eur per month (less than 3 eur per working day). That is nowhere near to sustain me, not even another UX guy. It is a known fact that it's very hard to sell apps on Google Play and actually make a living out of those - only perhaps games earn enough to sustain the dev team.

@mvysny

This comment has been minimized.

Copy link
Owner

mvysny commented Aug 9, 2017

Also see #768. In short, Android API sucks, developing for Android sucks, and the money is laughable. Even paid apps are nothing but a hobby, so it's expected to see mediocre quality.

@Akinori0713

This comment has been minimized.

Copy link
Author

Akinori0713 commented Aug 9, 2017

@mvysny

This comment has been minimized.

Copy link
Owner

mvysny commented Aug 9, 2017

Thank you very much for your contribution, it's highly appreciated! I also hope that after the money is raised, I'll be able to pay an UX guy.

Actually, the Kanjipad Extension also uses a modified version of KanjiVG, so the results should be at least similar. Hard to say; perhaps Shirabe Jisho modified the engine in a different way, or maybe they do some other tricks?

I am trying hard to think of a way on how to have the very same input going into Shirabe and into the Kanjipad, so that I could see the outputs and debug Kanjipad to find the bug, but I can't see any easy way to do so...

@mvysny

This comment has been minimized.

Copy link
Owner

mvysny commented Aug 9, 2017

Also the +-2 strokes option auto-turns off after 10 strokes since then the algorithm starts to get really slow. But perhaps Shirabe doesn't do that? Does Shirabe return e.g. 12-stroked kanjis even when 13 strokes are drawn?

@Akinori0713

This comment has been minimized.

Copy link
Author

Akinori0713 commented Aug 9, 2017

@mvysny

This comment has been minimized.

Copy link
Owner

mvysny commented Aug 9, 2017

In my eyes that implies that Shirabe is quite forgiving on stroke count even on more complex kanjis and thus uses the +-2 strokes even for complex kanjis. I will thus make Aedict to honor the +-2 strokes setting for a complex kanjis as well; after the next Aedict 3.42 is released, let us observe whether the situation improves or not.

@mvysny mvysny added bug and removed user feedback needed labels Aug 9, 2017

@Akinori0713

This comment has been minimized.

Copy link
Author

Akinori0713 commented Aug 9, 2017

@mvysny

This comment has been minimized.

Copy link
Owner

mvysny commented Sep 12, 2017

Thanks! Fixed in 3.42. To my horror, I realized that the +-2 strokes checkbox was completely ignored in 3.41, so it should work properly now.

@mvysny mvysny closed this Sep 12, 2017

@Akinori0713

This comment has been minimized.

Copy link
Author

Akinori0713 commented Sep 12, 2017

@mvysny

This comment has been minimized.

Copy link
Owner

mvysny commented Sep 12, 2017

Actually you will need to wait until 3.42 is out; it's not yet out since I want to fix several more bugs ;)
Also, +-2 strokes is very slow for kanjis with 12 and more strokes, thus only kanjis with +-1 stroke will be searched for kanjis with 12 or more strokes even though "+-2 strokes" checkbox is checked. Let's check whether this is enough for more accurate results or whether I really need to perform a full +-2 strokes search.

@Akinori0713

This comment has been minimized.

Copy link
Author

Akinori0713 commented Sep 12, 2017

@Akinori0713

This comment has been minimized.

Copy link
Author

Akinori0713 commented Sep 20, 2017

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
You can’t perform that action at this time.