Beta 5 Bug Report / Feedback Thread

[Fleminator]

Its getting worse...Soft resets occur when I am in Manage Flashcards/Settings.../Last Reviewed and I select ANY date. So that feature is not working anymore for me. I went to FileZ, deleted all the PLEN preferences (but left PLDE), re-installed my keyfile for Beta5, but I still have the same problem. Soft Reset EVERYTIME I leave the program, and then on certain operations. (I haven't check everything that causes a reset, but filter flashcards in Manage Flashcards window, just not the "Settings.../Advanced..." search window)...other search items appear to be working, just not those involving the calendar

Its also really annoying that the default dates are all circa 1970 instead of the current date...I'm sure this is already on your list of things to do.

 

[mikelove]

daniel123 - the input fields on the Edit Card screen will still only look like 1-line fields, but they'll have scroll arrows so that you can edit text that's too long to fit on one line; unfortunately we can't make them as big as the ones in 1.0 since we added separate fields for simplified/traditional character headwords. There's a glitch with text buffering in that pf version which could cause some issues with long imports, we think we've fixed it now though.

Fleminator - we've found/fixed a stack corruption issue with stroke order diagrams that could potentially cause crashes even when exiting the main screen (actually observed that happening several times on a T5) - if you specifically avoid using stroke order diagrams at all during a Pleco session, does that improve matters any? My question about where you were looking at the entries was actually relating to the crash, not the duplication (sorry I wasn't clearer on that) - did it crash when you exited the Edit Entry screen or the main dictionary screen?

There are lots of reasons why that repetition-spacing session might have come up with a different number of cards - duplicates are certainly a big part, both cards that were duplicated within that 2363 and cards that were duplicated elsewhere where the other version had been reviewed more recently (and hence its score took over). If you allow duplicates in the import, you should then get something pretty close to that 2363.

If you're using audio with those self-scored sessions that would explain why they're dying after about 40 cards, there was a nasty memory leak in the audio decoder (not even our fault, it was in the original open-source code) which would typically exhaust a Palm's available free memory after 40 or 50 plays. A "grace"/spread-out-to-be-reviewed-cards and fullscreen handwriting in free-answer are both good ideas, but not possible in 2.0 at this point; temporarily decreasing the number of points per day for repetition spacing might be a good workaround for the former, since it wouldn't affect the card scores but would let you easily put off some cards until later.

Not sure about this crash with date-based card filtering - could you send me a copy of your flashcard file in its current state? Perhaps there's still some corruption creeping in there, though it seems more likely the system is being destabilized by one of these other bugs. And yes, we do need to fix those default date settings.

llammamama - interesting, we'd probably go for the simpler redistribution approach at least initially; a calendar is a good idea but again not something we can consider for 2.0.

 

[The Duelist]

Hi Mike

I've reorganized my Imported Flashcards into Categories/Sub-categories.
I really like this new feature 8)

Crashes/Bugs
Platform: Palm T5
PlecoDict: 2b5(pf-vfs)

I'm finding the New Flashcard System Very Unstable.

Free Answer Session Seem Quite Good (Stable)
But Most Other Types of Test Sessions Crash after about 3 Cards.

I've Set-up/Tried Several Testing Profiles (See below),
with both Frequency Adjusted & Repetition Spaced Card Selection.
Also running the same test seems to Always select the same cards, (after crashing).

Examples Of Flashcard Testing Profiles:
Test Type: Multiple Choice
Show: Head+Pron(+Audio)
Ask For: Definition

Test Type: Self-scored
Show: Head+Pron(+Audio)

Test Type: Multiple Choice
Show: Pron+Defn(+Audio)
Ask For: Headword

Test Type: Self-scored
Show: Head+(+Audio)

Test Type: Stoke Order

Test Type: Tone Practice

 

[mikelove]

Yeah, from the number of flashcard files we've seen with 20 or 30 lists for a single textbook we think that grouping categories is going to be a very popular new feature

If you try those sessions without audio, do they then work without crashing? Could be related to the audio memory leak I mentioned before - the extra memory usage of a large flashcard pool like yours could certainly make the system run out of memory faster. The same-cards thing is a known bug with frequency-adjusted tests in that pf-vfs build (think I mentioned it in one of my posts in that thread) - do repetition-spaced cards show up in the same order too?

 

[ipsi]

I assure you, grouping flashcards is already popular with me!

One thing, though: If I have in the XML something like

<categories>
    <category name="NPCR" />
    <category name="NPCR/NPCR Book 1" />
    <category name="NPCR/NPCR Book 1/NPCR 1" />
</categories>

I get three top level categories called NPCR, one with a child, one with a child and a grandchild, and one without either. Is this intended behaviour or a bug?

 

[The Duelist]

If you try those sessions without audio, do they then work without crashing? Could be related to the audio memory leak I mentioned before - the extra memory usage of a large flashcard pool like yours could certainly make the system run out of memory faster. The same-cards thing is a known bug with frequency-adjusted tests in that pf-vfs build (think I mentioned it in one of my posts in that thread) - do repetition-spaced cards show up in the same order too?

Hi Mike

I did some Testing with the Following Profile:
Platform: Palm T5
PlecoDict: 2b5(pf.vfs)

Test Type: Multiple-choice
Show: Head+Pron (also +Audio)
Ask For: Defn

Card Selection: Repetition Spaced
This Testing Profile was OK (Stable). (with & without Audio)
Stopping a Session & Starting a new one gives different cards Each time(no repeated cards).
However, once there are no more cards to select, starting a New Session Always Crashes.

Card Selection: Frequency Adjusted
I am able to Run a Complete Session with 5 Cards. (with & without Audio)
Repeating Tests Always Tests the Same cards (already known).
Trying to run a session with 10 cards Always crashes at about Card 3. (with & without Audio)

 

[weight]

Hi Mike,

1) I have two TX's. I took the one that that I was originally having problems with; did a factory reset, reinstalled Pleco and restored my flashcards; afterwards I still got a soft reset when exporting. I took the other TX did the same thing and had no problems exporting. Except for FileZ, there was no other third-party software installed.

weight/Fleminator - you're both quite sure you have enough file storage space for these exported files, right? They can get pretty big, 5 MB or more. Also, could you each give me a list of all of the other third-party software you've installed on your TX? Since we have yet to observe this problem with the converter on a single system here, TX or otherwise, I'm starting to think it might be some sort of a software conflict.

2) The exporter strips slashes "/" out of my flashcards, can/will this be fixed?

3) I have always seen the zhuyin symbol "一" (as in ㄋ一ˇ, 你 ) as a horizontal line, the same as the Chinese character "一" (as in 一二三）. Why is it a vertical line "|" in Pleco?

 

[mikelove]

ipsi - I just tried importing that bit of XML and I'm not seeing those same problems; in fact there's a uniqueness constraint in the database that should make it impossible to have two categories with the same name + location. Might be some sort of display glitch - if you exit and reopen, do the extra categories still appear? Do you see multiple categories with that name in the actual SQL database?

The Duelist - the crash with no cards to study matches up with a bunch of other bug reports we've been getting, but I'm not sure why going from 5 cards to 10 would affect frequency-adjusted sessions; could be the same bug that's causing them to always show up in order, actually. Anyway since we've got a copy of your flashcard file we'll see if we can reproduce it.

weight - interesting to hear that you've had the same TX experience that we have; definitely sounds like this faulty hardware issue striking again. (does that TX also tend to get into a lot of reset loops?) Are slashes being stripped from your 1.0 or 2.0 exports? Everywhere (definitions too) or just some specific places like headwords or category names?

I'm not a Zhuyin expert, but the version of the alphabet listed at http://en.wikipedia.org/wiki/Bopomofo uses a vertical bar, as does the conversion table in the back of the ABC dictionary, so it seems like that's more standard - where did you study Zhuyin?

 

[Fleminator]

Fleminator -
if you specifically avoid using stroke order diagrams at all during a Pleco session, does that improve matters any?

It seems to have virtually no affect. Every now and then, I would click for Character Info...so I'll try to avoid that in the future, but I've certainly experienced resets that occured in sessions when I never looked at a stroke order diagram.

My question about where you were looking at the entries was actually relating to the crash, not the duplication (sorry I wasn't clearer on that) - did it crash when you exited the Edit Entry screen or the main dictionary screen?

Main Dictionary Screen, as well as a lot of other places

There are lots of reasons why that repetition-spacing session might have come up with a different number of cards - duplicates are certainly a big part, both cards that were duplicated within that 2363 and cards that were duplicated elsewhere where the other version had been reviewed more recently (and hence its score took over). If you allow duplicates in the import, you should then get something pretty close to that 2363.

duplicates couldn't possibly account for the discrepancy. I had 4006 total cards. Beta 5 imported 3764.
4006-3764=242 possible duplicates that were eliminated on import.
242+622 cards appearing<2363

To avoid actually studying for half an hour, I mapped out how I think things should transfer.
Points / # cards in PB5 / # of cards that should have mapped (including about 242 total duplicates)/Ranks/days assigned to ranks in 1.03
0-100/ 141cards (B5) /165 cards (1.03) /Rank 1-6 /(0 days)
101-400/ 11cards /75 cards /Ranks 7-9 /(1-4 days)
401-800/ 35cards /108 cards /Ranks 10 /(8 days)
801-1600 / 137cards / 587 cards /Ranks 11-13 /(9-14 days)
1601-3200 / 303cards / 1171 cards /Ranks 14-17 /(17-30 days)
3201-6400 / 1112cards / 734 cards /Ranks 18-19 /(45-60 days)
6401-12800/ 1039cards /816 cards /Ranks 20-21 /(75-90 days)
12801-204800/ 984cards /350 cards /Ranks 22-23 /(180-365 days)
204801 / 2 / /0
total /3764 / 4006

It really wouldn't bother me if you just say, "conversion is a complicated thing. If It took into account each cards history and bumped the score up"...but if each Rank's "Days" are supposed to map directly to points, then mine had a strange outcome.

Not using audio has helped my flash-card sessions immensely (by stopping soft-resets)

 

[mikelove]

Must be another bug then; there are a couple of issues with forms not cleaning up completely on exit that we're still tracking down, hopefully once we release another beta with the crashing bugs we know about fixed it'll become clearer what's causing the ones we still can't explain.

I'd be perfectly happy to dismiss this as a conversion glitch, but from those card counts it actually sounds like the bug might be in 1.0 instead of 2.0 - given that you have less than 1000 cards with spacings of under two weeks (even going by your 1.0.3 numbers), it seems unlikely that you'd have 2363 coming up after a two-week break.

 

[Fleminator]

Well, maybe it was more like a three week break
After looking at what was going on in 1.3...I don't think its a bug there. Probably just poor planning on my part that would slam with 2500+ cards after three weeks (but looking at it, started to simmer down towards the end)...I still don't have all my questions resolved regarding that transfer, but as long as I'm not slammed in 1 to 4 months with thousands of cards that I should have been reviewing closer to every two weeks, I appreciate the redistribution of my cards.

Thanks

 

[mikelove]

Hmm... what happens if you set the date forward on your system by a month or two, do more cards come out then or do you still see just a few hundred? That would at least establish that the cards are there, they've just been spaced out a bit more. (you could also try temporarily increasing the points per day to say 400 or so, that would compress the timeframe and hence show you a lot more cards)

 

[llammamama]

I've also seen some pretty strange behavior using "frequency adjust" "endless" "self-scored", and a relatively small set of about 20 cards. "Don't repeat cards in session" and "only change score once per day" are NOT checked.

I would hope that the program would just keep giving me cards ajusted to their changing frequency as I proceed to review them, however it seems to get unhappy once I make one pass through the set of cards, after which it begins to give me 3 or 4 of the cards I scored Best on over and over again. It also seems to be giving me cards with higher scores more often than I would expect.

The purpose of this is for me to get a list of vocabulary words into my head temporarily using the frequency ajust endless review (on a different score file), before I put them into my repitition spaced categories. Unfortunately at the moment it doesn't seem possible to use frequency adjusted for this purpose.

 

[Fleminator]

I know I need to stop asking for more features, but...

It would be nice if Pleco asked for a final review if a session is manually ended early (i.e. I push the Red "circle-x")...it already asks if I want to save the session, if the answer to that were no, it could proceed to present me with the cards I missed during that session...is that something that can be hammered out in a short amount of time?

I adjusted the date to Sept 30th (the expiration date, I take it) and it present me with 2400~ cards...I'll take that as answer enough. Thanks Pleco, for re-spacing my cards, and making catch up after slacking so much easier.

I tried emailing you my flash-card file, but its a little big (especially for my slow internet)

 

[ipsi]

Mike - Not sure. Will have to play around with it some more. I'm pretty sure that did occur the first time I imported some stuff.

Another minor bug:

If you have a flashcard of the form 一点(儿)[-點(兒)], it actually displays as 一点(儿)[-點-兒-]. Interestingly, I had exactly the same bug in my own conversion/merge program . Originally thought I'd fucked up and would have to fix them all again...

 

[plecoUser]

Flashcards and task manager

I've noticed a few bugs regarding flashcard use and how my PDA (ipaq rx1950, WM5) handles task switching:

* Once I've started a flash card session, it appears to live forever in memory until I manually 'stop' it via the windows 'running programs' menu.
* The 'Resume flashcard session' prompt doesn't work - it takes me to the last flashcard I viewed (correct) but the navigation at the bottom of the screen remains visible (it should disappear) and the 'tick' and 'cross' buttons on the flash card screen won't work.
* If I'm in the middle of a pleco session (including flashcard session) and I go open up some other app in WM5 (e.g., solitaire) - the pleco dict menu items remain at the bottom of the screen and don't get hidden. It's as though pleco completely takes over everything in my PDA and I can't use other programs at the same time without ending up with display issues of the navigation at the bottom of the screen.

Hope this helps!

 

[sjhuz01]

I think I fall in the category of people who are tech-savvy enough to install it, but not enough so to fix it when it all went to crap... Hopefully a little "standard user" feedback is useful too? I uninstalled after about 10 minutes following multiple resets, issues w/installing (it couldn't find the fonts), and (once I finally got it running & activated) several problems w/memory errors and crashing while it attempted to search.

It may just be that my 5-y.o. Palm Tungsten E2 with roughly 2MB free of the 32MB internal memory didn't have enough power to pull things off the 1GB SD card quickly enough. It couldn't seem to handle simple searches, much less cycle through multiple dictionaries.

I liked what I saw of the revised preferences menu - dictionary mgt, etc. Didn't get a chance to work w/the new flashcard options.

2 user interface issues:
- After clicking the "input" button that apparently cycles through handwriting, radical & keyboard, the button completely disappears after clicking it once & you have to use the menu or a separate toolbar button to bring back the different input options. Maybe it would've become more obvious over time, but there's no way to tell what mode it cycles to - the button's just gone.

Once I figured out how to get the input options back, the letters under the different keyboard options appeared squished into the same box (ie 2 letters together w/o lines separating them). The letters still worked independently - just didn't look right. This happened w/both Eng & PinYin keyboards - Zhuyin seemed to be ok, but I dunno.

Not sure why, but when attempting to look up an English word ("Zimbabwe" just to see which entries were included w/demo dictionaries), the "英" became highlighted in black & I couldn't change dictionaries or languages for several seconds. Maybe this was due to using one of the "demo" dictionaries? After a while it would unfreeze & I could try to highlight or search for something else, but that's also when it seemed to crash. Happened 2-3 times before I gave up.

Still looking forward to 2.0, but if the final release has the same issues w/speed & stability on a Tungsten E2, I'll have to stick w/1.0 (which it took me a while to reinstall & reconfigure how I had it ... had to get delete it to make room for the Beta...) Would gladly spring for iPhone 3G version if it's a "full" and stable version, and once they offer the darn thing in Taiwan...

Still very happy w/1.0

 

[mikelove]

llammamama - even with "don't repeat cards in session", in order to avoid silly situations like a card coming up twice in a row the system uses a "penalty box" to temporarily take cards out of circulation for a little while after you've just reviewed them; the size of that penalty box is supposed to change based on the number of cards available, but with small batches like that it sounds like it's being made too big - should probably only be one or two cards.

Fleminator - already done, actually; that's been on the list for a while since it's the only way to review incorrect cards in an Endless session. And thanks for the file.

ipsi - interesting, it's running pretty much the same code that the regular dictionaries do to split those so maybe something's different in the input text.

plecoUser - The issue with resume not hiding the menu bar has been found/fixed, but we haven't seen these other two bugs yet - what do you mean by "live forever in memory," does the session stay there even if you tap on the button to exit it? Where are you seeing that it's still around? With the menu item takeover, try going into Preferences and disabling "From Taskbar Icon" in the "External" section - does that help?

sjhuz01- the E2 is kind of a borderline system for 2.0, but the free internal storage space shouldn't actually matter; data pulled off of the SD card is stored in a different part of the Palm's memory. Is there any chance that the card might be running low on free space too, or that the files might not have been copied over to it completely? Does it help matters any if you delete all of the demo dictionaries? Does the input button disappear but leave a blank space where it's supposed to be, or does the panel seem to get rearranged with no room for it at all? The inverted "英" came up because the software switched to full-text search mode when it couldn't find an entry for "Zimbabwe" any other way - you can turn that full-text fallback off in preferences (it'll be disabled by default in the next beta since it's kind of confusing).

 

[The Duelist]

Hi Mike

I would like to make a suggestion for the Multiple-choice(MC) User-Interface(UI).

Could the MC Buttons ie Confirm/Cancel be the same as the Self Scoring Buttons.
Confirm=Tick
Cancel=Cross (Then move back to reselect an option)
I think this would make the UI more cohesive.

Maybe An Auto/Immediate Score Option for MC,
Would allow this Testing Mode to Flow more naturally.

What do you think?
The Duelist

 

[ipsi]

Not sure about that. I double-checked the flashcard entry, and the brackets were indeed in place. Happens when viewing a flashcard in a session or when viewing a flashcard in the manage flashcards screen. Changing to traditional produces the expected result, but the brackets are still missing inside the []. Odd.