19 Responses for "beta tests"

1. 
   ibmer August 25th, 2010 at 7:10 pm

   Hello Benoit,
   Using the “drawing pad”, try to write the kanji corresponding to the number “9″ following the correct kanji stroke order and you will notice that the kanji does not even show on the list! As this is a very simple kani (2 strokes), if you start by the second stroke, then the kanji appears in the list… in third position. This is unfortunately not the only example for which a basic kanji is not recognized but this example is quite significant I guess! ;O)

   Browsing many of the beta versions, it looks like the problem has appeared right after the release “3.2.0.24″, if this info can, somehow, help.

   I hope that you will be able to locate and fix the issue quickly… ;-p
   Cheers.
   Y.

2. 
   benwulf August 28th, 2010 at 1:10 pm

   I see it, I’ll fix it for the next version to go to the appstore after the one currently being approved (so after 3.4.3).

3. 
   ibmer August 28th, 2010 at 1:46 pm

   Salut Benoit,

   Thanks for your prompt reply and many thanks in advance for the fix! I wish the Apple developers could be as reactive as you are! )

   In the other hand, Apple’s inertia on some development areas where expected ROI is very low shouldn’t be seen “that bad” from your perspective, right ? ) And honestly, I don’t believe that kanji recognition will be implemented on the iphone in a near or even a middle future so you can continue to unleash your creativity and skills for the BIG pleasure of the japanese learners community I’m part of!

   All the best, a final thank you for your creativity and cheers,
   Y.

4. 
   hex September 8th, 2010 at 10:05 am

   Hello Benoit,
   I’m just wondering, is there a reason why kanji frequency:1727, Japanese readings On: tou Kun: kara has a short definition in English T’ang; China but in Spanish apart from that also adds the definitions mentira; exageracion; tonteria; de pronto. was that intentional or a mistake?

5. 
   benwulf September 8th, 2010 at 8:00 pm

   Hello Hex,
   I don’t know really, the content is the one from the jmdict dictionary. It seems that the person who made the spanish entry entered more meanings than those who entered the other languages.

   I did check with my wife who is native japanese and she didn’t think “lies” “exaggeration” (my interpretation of the spanish, as I don’t speak spanish I may be wrong) really applied to this kanji except maybe as they relate to negative prejudice against chinese people in Japan.
   Best regards
   Benoit

6. 
   Singaporian October 13th, 2010 at 5:24 pm

   Hi Benwulf;
   Just for your information, the last beta release has a severe bug (closes as soon as you click on the upper menu bar) and the previous one hasn’t correct the stroke order bug mentioned by “ibmer”… I’ve used you application thanks to my gf who’s studying japanese and would love to test the upcoming features via your beta program so I’ll be looking forward on next WE as you seem to release the beta on sundays Gambatte ne.

7. 
   benwulf October 23rd, 2010 at 7:56 am

   Hello Singaporian,
   I was in a business trip for my day job the last 2 weeks so couldn’t work on the beta. I’ll post a new one this WE with the fixes. Not sure about the one for the stroke order, this will be in the next appstore version as mentioned but rebuilding the stroke order database is something I normally do only at those time. I have fixed the bug in the reference database though

8. 
   Yayoo March 19th, 2011 at 11:45 am

   ibmer
   August 25th, 2010 at 7:13 pm
   Hello Benoit,
   Using the “drawing pad”, if you try to write the kanji corresponding to the number “9? (strictly following the correct kanji stroke order of course), you will notice that the kanji does not even show on the list! As this is a very simple kanji (2 strokes), if you then start by the second stroke, the kanji appears in the list… in third position. This is unfortunately not the only example for which a basic kanji is not recognized but this example is quite significant I guess! ;O)
   Browsing many of the beta versions, it looks like the problem has appeared right after the release “3.2.0.24?, if this info can, somehow, help.
   I hope that you will be able to locate and fix the issue quickly… ;-p
   Cheers.
   Y.
   ____
   benwulf
   August 28th, 2010 at 1:08 pm
   Yes, I see it, I’ll fix it for the next version in the appstore (after 3.4.3 which was already sent)
   ____
   Hello Benoît,

   I’ve just bought your last version today (v3.5.0.15) and this MAJOR problem is still not corrected . Thanks in advance for the update!

   Try to draw “9″, the keystroke order is incorrect.

   Y.

9. 
   Yayoo March 19th, 2011 at 12:03 pm

   Salut Benoit,

   I’m just bought your app and got the dialog box “App crash detected” each time that I launch the app!

   Also, try to draw “9″ on the drawing pad while respecting the correct order : the kanji doesn’t appear on the list. This problem was not existing with the version 3.2.0.24! Sadly, many order kanji are now not recognized…

   Reading your reply to “ibmer”, this problem was supposed to be solved since the version 3.3.xx ?!?

   Cheers,

   -Yahia-

10. 
    Yayoo March 19th, 2011 at 12:19 pm

    Benoit,

    Sorry for having pushed 3 “messages” today… It will be my last one for a few week, sorry!

    This last remark might be interesting for you and also many other users!

    On the AppleStore, you have provided screen captures of some of the application screens. It we consider the 3rd screen capture (the drawing pad), it shows little red dots which define were each stroke has started en ended. In the application, the only little red dot are shown in the main window but not in the drawing pad…

    While this option is shown on the screen capture and would be very useful… why have you remove it on the app shipped on the Apple Store ?

    Many thanks in advance for all your efforts with the corrections and your great app!!

    Y.

11. 
    benwulf March 19th, 2011 at 3:07 pm

    Actually I removed it on the suggestion of others. What this was doing was showing explicitly which of the points in the stroke are taken into account in order to recognize the character.

12. 
    Yayoo March 19th, 2011 at 8:12 pm

    Hello Benoît,

    And thanks for your prompt answer!

    I see… Then, it would be REALLY nice if you add this USEFUL feature as an OPTION so that the ones who would like to see which little red points are taken into account! There’s room for an UIswitch in the current settings UIView and you can always use an UIScrollView… right ? ;O)

    Looking forward to reading your answers about my two other comments posted earlier today.

    Thanks in advance for your support and thank you very much for the interaction that you offer to your users… this is very appreciable (being an IOS developer myself at times, I know how “time consuming” and demanding development is!).

    Gambatte ne!

    Cheers,
    Y.

    PS. (May this piece of code give you some ideas for Shinkanji, please have a look : https://github.com/FLCLjp/iPhone-Zinnia. Note that the IOS libzinnia.a is not shipped but the IOS binary is quite easy to compile as the source is provided!

13. 
    benwulf March 20th, 2011 at 9:57 am

    Hello,
    first zinnia is what I use for the handwriting recognition and I have been using it since much earlier than the post by this person…
    regarding the reactivation of the “red dots” I can look into it but it will be lower priority than actual bugs, you can open a request in the bug tracker.
    As you guessed the challenge is more about where to put the option switch (there isn’t much place left and using a scrollview makes the UI more complicated).

    Best regards
    Benoit

14. 
    benwulf March 20th, 2011 at 10:04 am

    As mentioned I am looking into it. Part of the problem is that the Zinnia library which I am using has an algorithm which is quite complex and requires that the model characters be transformed before being used for recognition. There is a difficult to adjust parameter which modify the size of the database, I am trying to find what is the best value.
    Best regards
    Benoit

15. 
    Yayoo. March 20th, 2011 at 1:16 pm

    Hello Benoît,

    And thanks for your quick and sharp answers!

    As you have to focus and put priority on the bugs correction, I would like to report you that each time I launch “Shinkanji v3.5.0.15″ I get this dialog box :

    ” Crash detected
    detected that the previous sessions ended in a crash, removing all settings.

    YES NO ”

    Run on an : jailbroken iphone 3GS running IOS 4.0.2 (rebooted after install)

    I’m now using again the 3.2.024 version which is stable and makes me happy in regards of the handwriting model. If I remember well, that was the last version until you make the app universal to work on Ipad as well, right ?

    (Also, note that the old 4-level JLPT has been changed into a new 5-level JLPT.)

    Gambatte ne and thanks again for the interaction with your users!

    Yahia

16. 
    benwulf March 20th, 2011 at 1:27 pm

    Hello,
    regarding the handwriting model I just made changes to the db generation and it works for the kanji 9 and probably several others. The changes won’t be in the next version in the appstore v3.5.0.16 but in the next v3.5.1.x which will be sent to apple as soon as they accept v3.5.0.16
    Also I don’t understand your comment about 3.2.0.24 being the last universal app. Shinkanji is still a universal app and has been since I first released iPad compatibility. I just checked on the appStore (you had me worried) and I see the + indicating the universal condition next to it.

17. 
    Yayoo. March 20th, 2011 at 6:34 pm

    Hello Benoît,

    Wow, that was quick, thanks! All my apologies for having worried you :p.
    In fact, I thought that the bug came along, somehow, with the changes done to bring the ipad compatibility right after the v3.2.0.24 (Even if since, I have noticed that the handwriting model had changed… size has decreased!

    By the way, as you’ve been a major contributor to these projects, are you using modified versions of tomoe & stroke-editor to optimize your model or are you working with homemade code ?

    Many thanks for the good work and the interactivity!

    Y.

18. 
    benwulf March 20th, 2011 at 7:17 pm

    I did make some changes sometimes in version 3 to the generation of the database based on the Tomoe models. Mostly I only take into accounts those kanji which have + or – 3 strokes compared to the one being evaluated. When I did that I probably regenerated the database with the wrong set of data. In the old Tomoe models the 9 was badly rewritten and I had added a variant with the correct stroke order, so I think that during the later regeneration I must have taken the wrong model as inputs.

    Regarding my contribution to Tomoe I contributed udpates to the database to add correct stroke order variants where there were mistakes (I kept the mistaken one on the principle that if the original database writer made the mistake others would to and it would be interesting to recognize both variants) and a MacOS stroke editor. I don’t use tomoe’s code in shinkanji but zinnia just tomoe’s model
    Benoit

19. 
    Yayoo March 21st, 2011 at 12:53 am

    Great, I understand, thanks for the explanations!
    Have a nice week
    Cheers,
    Yahia