Beta 7 Bug Report Thread

[bglasow]

Touch Diamond, Windows Mobile 6.1, CE-Star

A few comments on installation with the Windows Mobile automated installer...

Installed on a laptop running Windows Vista Business. After confirming the install file, the first time it then gave an error saying it could not create c:\Program Files\Pleco. I realized I probably needed to right-click the .exe file and run as administrator. After doing this and allowing escalation privileges it started up okay.

After the files had been successfully extracted onto my laptop, the installer brought up a screen asking where to install them onto Windows Mobile. It gave two directory options:
Internal
Internal Storage

The Internal directory is just the phone memory, and next to the name it correctly showed 38MB free space (everywhere else in other Windows Mobile programs this particular directory is referred to as "\", but that is a minor point as I understood the name based on space available).

The Internal Storage is the correct name for the non-removable 4GB storage card within the Touch Diamond. However, the install screen displayed no figures about the space available next to the name of that directory. When I chose that directory to do the install, it simply popped up an error message saying Inadequate Space - Please choose another directory Or Delete some files and try again. This was inaccurate- that Internal Storage directory actually had 33GB free space. Eventually I gave up and manually installed the files which worked like a charm.

 

[johnh113]

Dear Mike,

I'm having problems with my flashcards.

It's probably because of my odd study habits. I have been studying Chinese for several years, and so have used many textbooks. I always put all of the vocabulary words from the textbooks into my flashcards, along with the HSK and various other flashcard lists. I'm sure I have billions of duplicates, but I happily restudy cards that I've studied before.

I have 27,254 cards. When I input them into Beta 7, I allowed duplicates, so I still have 27,254 cards. When I tried frequency adjusted (by mistake, I really don't care about frequency adjusted), I got a soft reset. When I tried repetition spacing, the little timer thingy spun for about 50 seconds and then I got a soft reset. This happened several times.

When I used far fewer cards, it worked fine.

If you tell me to just get rid of my duplicates, that's fine. I suspect I'm the only person that studies this way.

John

 

[mikelove]

ldolse - we actually just fixed that bang4ong1ting1 bug yesterday (found it thanks to an ambiguous entry prompt while importing someone's flashcard file) - the part of the Pinyin tone-to-number converter that detected mid-syllable vowels was for some reason missing 'o' in its list of vowels to check for.

With the multiple-choice bug, were you using the dictionary definition remapping feature in that session? If so, that's caused by a known bug (now fixed) with the list of dictionaries to remap to not being retained correctly in saved sessions. If you weren't using that then this is something new - was this using all cards as possible wrong answers or cards from a specific list of categories? (there wouldn't be an issue with those categories saving out incorrectly because the list of possible answers is generated before the session begins)

sych - looks like a change we made to get searching in custom dictionaries working correctly mixed up flashcard searches - found/fixed now. Not sure what's happening with that disk space error but it should be easy enough to check for... and you've contributed plenty, don't worry - glad you're liking 2.0.

Sarevok - interesting that it would happen in there too, we'll do some more testing with restoring sessions that are in the midst of being backtracked. The other bug sounds like it's probably related to that issue with saved session files not being deleted correctly, we'll see if we can reproduce it with that fixed. With the Mdict problem, this sounds like a consequence of our use of custom user interface controls for text display / input fields (which were necessary in order to work around the many many many bugs in 1.0 that were caused by Windows CE's horrible font rendering system) - our custom fields still respond to standard copy-text-to-clipboard messages, though, so a dictionary program that uses the clipboard (rather than attempting to copy text directly from the control) should work fine with them. Pleco's Instant Access system does this, and it would pretty much be the only way for another program to pull text out of tricky custom controls like Opera browser windows, so I imagine other dictionary programs might do it too. Which is a roundabout way of saying that if you try another dictionary program besides Mdict you might have better luck.

stephanhodges - thanks very much for these. Already responded by e-mail so I'll just put that in here for the sake of others reading:

The slowness / refusal to advance probably has to do with the fact that you were opening a UTF-8 document in UTF-16 - it may look like a perfectly innocent field of Chinese characters, but there are actually a whole lot of invisible / invalid characters being skipped over when those are drawn, and since opening files in the wrong encoding isn't really something we expect people to ever deliberately want to do we haven't taken the time to optimize the software for those cases.

(you've basically got a situation where a string of 8-bit values is being interpreted as 16-bit character codes - almost half of those 16-bit character codes in Unicode describe Chinese characters, so just randomly you're going to stumble onto a lot of Chinese character codes, but they have absolutely no relationship to the Chinese characters in the original text)

The difference between the left/right arrows is that the left arrows move the left edge of the selection while the right arrows move the right edge. So if you tap on the right advance arrow, it searches all dictionaries for the highlighted word with that extra character added; if it doesn't find it, it removes characters from the front of the search query until it does find something it can use.

The menu isn't accessible in the actual reader document because there's a whole separate system that manages text selection for reader documents - getting that working together with the one for dictionary definitions would be really complicated and likely introduce a bunch of new bugs, so not something we can think about right now.

Good idea on toggling that definition bar, we're planning to make the reader interface a lot more customizable in 2.1 and that would definitely fit with that. The bookmarks/notes ideas are also promising, certainly there's a lot more we can do with the bookmark system. Though we'd probably need to put the bookmarks in a SQLite database if we were going to add large amounts of user-created data like notes, and that might not ever happen on Palm since adding another SQLite database to the mix is going to strain the system even more than the existing ones do.

The magnify command isn't supposed to work at all with non-Chinese text, it looks like the port over of Stylus Actions to the document reader lost the Chinese character filter but we'll add that back in so you won't see a black box. Not sure what the problem is with Char Info - to be honest I think adding Stylus Actions to the reader may have been a mistake, it's turning out to be a lot buggier than I'd expected, so we may end up either taking it out entirely or disabling it by default and re-enabling it with a checkbox (coupled with a warning message about how buggy it is) - way too big a new feature to have added this late in the game.

The "de" in ABC is correct, you can use the arrows to scroll down to the other entries for that character but "d" is a perfectly valid one (which you can find in the main dictionary too).

We'll take a look at the resizer bar, might be a landscape display thing. Dictionary history list box size is indeed incorrect, looks like we forgot to specify different sizes for it on Palm and WM (Palm has a slightly larger border) so when we fixed it on WM we ended up screwing it up on Palm.

sych again - interesting; ipsi reported problems with IA crashing Blazer earlier but I thought in his case at least we'd fixed them. (ipsi?) With crashing on audio playback this may just be a case of the system running out of memory (audio playback being considerably more memory-intensive than anything else in IA) - if we can't get it working we can probably at least check for programs that it generally doesn't work with and disable the audio button when launched from them to prevent crashes.

estudiando/sych - not sure what's going on there, the keystrokes should still be added to the Palm OS event queue either way (and forwarded to Pleco once the search completes, as they did in 1.0) - maybe something else is intercepting them, we'll take a look.

bglasow - interesting, that administrator issue didn't come up in our testing on Vista - must be a security configuration thing. There's an option in NSIS to put up that escalation prompt automatically but I thought it was already enabled... the 4GB issue sounds like an overflow problem, I think we're still storing that value in a 32-bit variable instead of a 64-bit one like we should be - should be easy enough to fix anyway since the getdiskfreespace function definitely returns a 64-bit value.

johnh113 - I think the problem there might be simply the number of cards, actually, rather than the duplicated ones specifically. What type of Palm are you using again? If you limit the session to just a subset of your available categories, that might reduce the memory requirement and allow the session to start up more smoothly.

 

[johnh113]

Dear Mike,

Yes, you're right. I didn't think the duplicates were that problem except that it made for 10's of thousands of cards. When I use a subset of cards everything works fine, except even then the flashcards are really slow to get started, I guess because of the slowness of mysql versus whatever you used in Pleco 1 which starts almost instantly even with all my cards.

I'm using a Palm Treo 680 with no other installed programs other than what came with the phone. Everything installed on the card except what you said had to be internal. Nothing running in the background.

John

 

[estudiando]

Additional info on the keystroke input problem:

When the dictionary is searching, the keyboard feedback (the key turns dark when you tap) is not there or when when you use graffiti, you don't see the trace being drawn. When the dictionary is done searching, some of the keystrokes you thought you entered are missing from the input line.

 

[ldolse]

I wasn't remapping cards to a new dictionary on the fly, but I was pulling my answer options from session categories instead of all categories.

 

[ipsi]

Sych,

IA from Blazer works just fine for me. Not sure what's up with it not working for you. Have you set the external encoding to GB2312/Big5? I assume you're using CJKOS?

I also got the audio-crash problem. A bit strange that it will happily play audio four or five times, then only crash when you hit 'done'.

 

[sych]

ipsi: yes, encoding settings in Pleco are set to GB2312, as are the encoding and display options in CJKOS. After playing with it some more, I've found it doesn't crash absolutely every time, but it's been about 7 crashes out of 8 times using IA in Blazer, so far.

 

[johnh113]

Dear Mike,

I decided I'd try to delete my flashcards and reimport them skipping duplicates. I went to PlecoFlash, Manage Flashcards, More and clicked on Remove all cats. It went through all my cards but all the cats were still there. I clicked Delete all... in order to delete my flashcards. Then tried again to delete my cats. No success deleting cats. Tried Remove and that succeeded in removing categories but it would have taken way too long to delete them all one by one. I removed about 10 before giving up. Then I reimported with Skip duplicates but keep cats. This time I had 15253 cards instead of 27254. (It doesn't seem reasonable to me that I would have 15253 unique cards since I get my cards from the PCR and HSK lists on the Pleco site, and from my elementary textbooks, which all seem to have about the same 4000 or so words.) Anyway, I started a new session and it soft reset immediately. I tried again and it again soft reset immediately. I tried a subset of cards and it worked fine. Is there someway I can delete all of my flashcards and categories and everything and try a fresh import? Is there a flashcard file I delete? Or should I delete Pleco2 and reinstall?

John

 

[ldolse]

Johnh113, I assume you used the new export utility? That would have kept all your score data too. Might be interesting to try doing a Vanilla export from 1.0 (old export, lose score data) and see what happens when you import that.

You'll still get quite a few dupes based on traditional variants I imagine. The current algorithm is cautious about deciding dupes, so if there are differing traditional variants with the same simplified character it will keep both cards. Same if the Pinyin is slightly different.


New weird issue to report:
I did the following search:
@字頭

This came up with several duplicated search results in ABC - there were 5 total results returned, but two were duplicate, so there were only three unique entries returned.


Lastly - comment on the stylus actions in the Reader - I find that it works plenty well, would rather it just be left as is, you can always clean it up in a later dot release. I can't even get it to reproduce the bugs mentioned earlier, but maybe it's only an issue on Palm.

 

[radioman]

Not sure where this one goes, but just a note that under the Character Info, when bringing up Compounds, it takes awhile.

- under the Axim, it was almost instantaneous on beta 6.
- under version 1 Pleco for Palm it took a few seconds sometimes, maybe 5 or 8 seconds.
- with beta 7, it sometimes takes me 15 seconds to a minute (or more??). Just really long.

Not sure if this is just me, something known and being fixed, or just the way it is.

I do have every dictionary I have been given access to running on the Palm, so maybe that has something to do with it.

The compounds function very useful (especially as it is not crashing...) and much better to have than not, just not sure if the slowdown was a result of the new version, or a configuration issue on my side.

 

[ipsi]

Sych: That's really bizzare. I'm not getting them at all, despite having pTunes running in the background with ChatterEmail connected to a GMail IMAP account via IDLE... Along with a couple of other programs. What sort of stuff do you have installed? It's possible there's just not enough memory free to be able to use it? But that shouldn't be the case if P2 can start at all... You have it localised, according to your sig. Does it happen if you remove the localisation? Does it make a difference selecting Chinese or English words?

Quick sanity check: You're definitely running Beta 7, right? Under Help/About, it should say "version 2.0b7".

 

[ipsi]

John,

If you're using a Palm OS device, probably the easiest way is to open up PlecoMover, which will allow you move or delete the various flashcard databases. The one you're looking for is called PlecoFlash. Ignore the others. There should be at least three options under "Move From", unless you don't have an SD Card. Select each of them in turn until you see a file called PlecoFlash listed. Then, if you're sure there's nothing you want in it, hit delete. Otherwise, mark it as backup and Pleco will ignore it, allowing you to start from scratch.

If you don't see a program called that, then install it. It should have come with the Beta 7 download, and is in the folder called "palmfilemover"

I'm not sure what to do if you're using a Windows Mobile device - I suspect someone else should be able to help you with that though.

 

[sych]

Hi ipsi:

Sanity Check: Yes, I'm sane. 2.0b7.
Free Memory: 19.9MB of 69M (according to "Delete" in Apps), 20M Internal Memory free according to FileZ.

I have just tried:
1. Turned off the Localiser
2. Reset Pleco preferences to defaults
3. Re-set my IA key
4. Re-set external encoding to GB
5. Rebooted
6. Opened Blazer, selected some text (tried "2008" first, so it's not Chinese chars), hit IA button, and got nothing.. (though no crash).
7. In Pleco, set Extraction Method to "Clipboard", came back to Blazer, selected some Chinese text, pressed IA, and crashed.
8. After the crash, came back to Blazer, selected some English test - no crash! Got a definition, too. Selected Chinese... crashed.... hmmm, ipsi, maybe you're on to something...
9. Came back, set Extraction to "Experimental", went back to Blazer, selected English text ("BBC") & pressed IA, no problem. Selected "中文", pressed IA, no problem. Selected "中文网" -- crashed. Maybe the issue is related to the length of the string?
10. Text "BBCChinese.com" -- no problem. "中文" -- again, no problem. "格林尼治" -- crashed.
11. Text "Civil Aviation Authority" -- no problem. A whole paragraph of English text; also no problem.
12. Text "西藏" - no problem, until I tried to change to another dictionary in the IA popup -- THEN it crashed.
13. On the hunch that the whole problem is to do with changing dictionaries, I changed my dict search order to Adso at the top (replacing Oxford), as Adso has an entry for "格林尼治", whilst Oxford does not. Strangely, using IA this time, selecting "格林尼治" got me an entry from Guifan rather than Adso, but didn't produce a crash. When tapped the icon to change dictionaries, THEN it crashed.

At this point I went to get some lunch and stopped writing down everything I was doing, but on the suspicion that the problem was related to searching certain dictionaries, or switching between dictionaries, I started playing around with changing dictionary search order - and disabling certain dictionaries in Pleco, and then trying again. However, the problem still occurs with just about any set of dictionary settings, and unfortunately I have not been able to find any pattern to it.
In case I might have had some corrupted dictionary files, I marked my User Dict as "backup" so it wouldn't be opened, and copied new dictionary files to my SD card. (So as the newly copied files would occupy different sectors on my card, I renamed the existing files rather than deleting/overwriting.) I also removed (renamed) the Tuttle database file so that it wouldn't be opened (as it is the newest addition, so possibly suspect). None of this helped, so I also tried reverting to the Beta 6 dictionary files, which also didn't fix anything.

So after all this I still can't point to what the problem is, but I hope it'll give Mike something to work with, and maybe a way for you (ipsi) to reproduce it on your T680.



More info:

Device Firmware Ver: MW01.50
Software: TREO680-1.04-ROW
Hardware: A
Carrier DB: 355

CJKOS Version: 4.63 (registered; latest ver, according to http://www.dyts.com)

Other Software:
IBM Java VM (don't use it much, don't think it has any reason to stay resident)
Missing Sync utilities (TimeCopy, disk mounter, and the Missing Sync conduit - though, AFAIK none of these stay resident, they're just conduits and utilities)
SuperMemo
DocsToGo
Google Maps
VolumeCare (installed, but not running (says "turned off" in options)
Standard Palm software, including bundled ver of PTunes (but not currently running)
VersaMail is installed and configured for an IMAP account, but not configured to check automatically.

 

[ipsi]

Very interesting. I can reproduce it here. Take the text 我是大学, highlight and try to use IA. It should explode. However, highlighting 罗马尼亚 (Romania) and using IA will work fine, as there's an entry for that. I guess it's just not handling text where it can't find a definition very well? Note that 中文网 is *only* in Adso (as part of 中文网页), for example, so if Adso wasn't enabled when you looked that up it would probably have crashed. In fact, it may have crashed anyway, if it has trouble looking up an exact match. Hmm...

It works fine from the SMS application.

Following the order you did things in seems to reproduce it for me too. That's very strange. I'll play around some more, but that's kinda odd.

Having said that, I'm now able to use IA to look up 中文网, whereas before I'm sure it caused a reset. But that was before it reset, so perhaps there's a memory issue involved? In so saying, I just tried to look up a random string from 红楼梦, which I have as a series of HTML files in Blazer (for before the document reader came about), and that crashed it. Crashed it every time, in fact, despite it being the exact same string. Odd. The string, by the way, is 今风尘碌碌 (jīn fēngchén lùlù)。 Tried three times, crashed three times. Tried looking up 今 first, but still crashed. Hell, it crashed after I looked up each of the words one-by-one.

Entered string into SMS App (default one that comes with the 680), and tried. Result: Success! I assume you use a different method of copying for Blazer and the SMS app? Tried looking it up from both the text-entry box and from a sent message - success both times. It only showed 今, as there's no such word 今风.

Looked it up again from Blazer - crash.

Entered it as a search string in Pleco before trying to look it up via IA - crash.

Tried again with just 今风尘 - crash.

It's worth noting that the HTML file is on the SD Card (accessing via file:///...), and that it's actually encoded with GB2312 (not UTF-8).

At this point, I'd say it's having trouble looking up a word where there's not an exact match, or where it's not a substring of a larger word. Very odd. Which doesn't explain why I can look up 中文网, and it have it show the Oxford def for 中文.

It also seems to be suffering from memory leaks. Played around with it for a while in Blazer (maybe 5 - 10 lookups) and it eventually crashed on 中文. And of course Blazer will happily trash my Plecoforums.com cookies if it crashes on this page. *sigh*.

But wait, now it crashed on 风尘, and that does exist. Brain hurts...

After further testing, I managed to get 风尘 working - not sure what went wrong there. Anyway, checked out 作者 - good. Checked out 作者自 - Got the ABC for 作者, but no crash, thus good. Tried 今风尘 - crash!

I think that's all for now, hopefully that's enough to replicate it Mike...

Oh, and by the way Mike, while I'm sure Sych would love to have this problem sorted, I've got a Treo Pro on order which I'm really hoping will fix most of these issues (and introduce lots and lots of other ones (I'm not hoping for that. Just being realistic)).

 

[Dan_78cj5]

Minor things compared to others on the forum...

two issues under categories. I am using CJKOS to input Chinese character names for my categories. My first category is 战役学-绪论 and it worked fine no issues, made a category and I can use and manipulate it fine. Input my vocab from this chapter/preface and assigned them all to the category, did a few flash-card drills and then moved on to chapter one. When I came back to the manage categories screen to create a second category, I again used CJKOS input keyboard to put in 战役学-一章 but now I get an error saying the title of a category cannot be "all spaces". After manipulating different versions of the name, restarting a few times, doing other functions and coming back to manage categroies I find I cannot input any category name that doesn't start with a roman numeral or english letter. Now, apparently it can have characters, but not start with them. Even though my first category is still working titled with a character at the begining.

In this process I also discovered a repeatable soft crash. If in the process of naming a category I open the CJKOS input keyboard and then hit either the key to input number or the key to input letters from within the CJKOS character input keyboard I get a soft reset as soon as I touch those keys (marked with an "A" and a "1" on the lower right corner of the CJKOS input keyboard.

All in all, very minor problems for me so far, very glad to have a clean looking stable 2.0 to use!

Thanks Mike!!!
Dan

 

[radioman]

A minor point, under the following scenario

In flashcards,

- I go to "character info" by clicking on the head character.

- I select compound (it takes me about 10 to 30 seconds to get a list).
- I see a variety of words, so I select one and "go to that word".
- a definition box (without borders) lays across the original flashcard session.

The first problem is there is no border, so it looks unfinished (in my opinion...)
The next issue is that if I then hit the "minus" flashcard button to continue (as well as say I did not get the original flashcard), then it returns to the flashcard session, but there are remnants of the popup definition that was provided.

It is also not clear on how you "resume" the flashcard session so I just went for what I could see at the time, the minus button. As well, my guess is that the problem would be exaserbated if the user sets the flashcard mode to "left handed" as the popup box would (I guess) cover both the main flashcard buttons.

###

 

[garysaville]

Pleco 2 beta 7 is working OK on my HTC Touch Diamond until I try to install the audio files. Then the program crashes during load time.
I have all of my files installed to /Pleco on the extended memory section of the HTC. The dictionaries are all working fine.
Any suggestions?


Thanks

Gary

 

[Mator]

Hey Mike,
Much more stability in beta 7 than 6 and most everything works well so far except for a few problems I'm having:

Crashes:
- In Prefs/Button Actions anytime I tap on "Other Button Actions.." to set them I get a soft reset
- Using IA I have the same problem as others - when I tap on Done after playing an audio file I get a soft reset. I also have no success with IA in the default Palm messaging program - nothing comes up when I press the IA button, but in all other programs it seems to work fine (except for audio problem).

Cosmetic:
- I agree with Radioman in that the compounds take a while to come up at times and when going to the selected word it does look very unfinished.
- In a flashcard session the clock (when on left) and the timer (when on right) overlap each other

One other thing that is annoying (and maybe not related to Pleco) is when using the HWR, it will hangup for several seconds at times in the middle of writing a character - maybe 10% of the time. I've only experienced it in beta 6 and 7 - any ideas? I also got a lot of hangups when putting in pinyin search in beta 6, but that seems to be gone in beta 7

Thanks, Mator

 

[sych]

ipsi:

I assume you use a different method of copying for Blazer and the SMS app?

No, "experimental" mode for both. It's worked for 1.x right up to B6. The only issue is that in MMSes, it's never worked the 1st time; only the 2nd and subsequent times. (doesn't seem to be able to grab the text the first time).