Beta 2 Bug Reporting

[daniu]

Hi!

Sorry about changing the topic:

Since I got a test next saturday I have to spend some time learning vocabulary these days ... As for some of the characters I am not 100% sure about the stroke order I am using PD2 more often ...

What I just cannot ignore is that PD2s character input misses strokes quite often. It is not just when I start righting outside the box (that was an issue in PD1 already) but also quite often it just seems to be slow ... and misses a stroke when I try to put in a character quickly. Really annoying!
Anybody else having that issue?

I also miss the character input directly in the dicionary entry box (full screen input) ... that is really convenient.

regards
Daniel

hmmm - when will Beta 3 come? ...

 

[sfrrr]

Mike--I'm one of those people who doesn't want to hear incorrect pronunciations. I remember words (tone, meaning) by hearing and repeating phrases over and over and linking them to the meaning and the character--a sort of right brain/left brain combo. I really think that a lesson on the order of ChinesePod's is much more helpful in this regard than trying to learn tones and accents from a dictionary. I study with a wonderful teacher, but now that most of my Chinese-speaking friends have moved away, listening to CPod several times a day drills stuff into both sides of my brain.

Sandra

Ogh, and I have a question about the tones presented in the flashcard files. A few words, most notably kan (to see or read), are given strange tones. For instance, I have always heard that kan4 is pronounced in...well, the fourth tone. So how come nit comes up in my PD2 flashcards as kan1? (I think I'm using the HSK set of flashcards. In Hong Kong, do they pronounce kan4 as kan1?

 

[ssaito]

hi pleco community

keep up the great work, mike & pleco team! pd2 is a big improvement. i like that you fixed where when typing in characters quickly the search function would stop searching before i finished entering characters. now search keeps searching until i stop typing.

my config: at&t 8525 wm 6 2g sd card (on which pd1 and pd2 are installed) and mobem cestar

i'd like to second Daniu's comments in his previous note. i experience all the same things.

also, i find that pd2 seems to forget that i loaded it with the keyfile. it remembers for a while, but when i shift away to another app then call pd2 back up, it sometimes forgets the keyfile. it also occasionally forgets where the dictionaries are.

Thanks,
Steve

 

[ipsi]

sfrrr - Apparently it can be pronounced as kān, though it does have a different meaning - check your dictionaries .

I don't mind hearing incorrect tones, I prefer that to no audio at all. Probably best to leave it on by default (I reckon you'll get less tech support emails about 'how do I turn this off' than 'there's a bug with your program: not all the audio plays').

 

[roddy]

The neutral tone thing is only going to be a problem when you have a word that is a) outside the 30,000+ ( :shock: ) word set of pre-recorded words and b) has a neutral tone in it. I'm going to guess that is a pretty small set, and not one anyone is going to be using every day. It's not going to affect frequent neutral tone words such as 哥哥, because they will have a specific word-level recording that will take the neutral tone into account.

It wouldn't be that easy to build up tone-correct words from single syllable blocks. You'd need to take into account tone changes - two third tones, bu's and yi's, and also the different pitches of neutral tone. I guess it could be done, but frankly I'm not sure it's worth the effort. It'll be easy to hear when you're getting an ad-hoc built-up word, at that point we'll just need to have a quick look at the pinyin and check what we're hearing is right.

 

[mikelove]

daniu - lots of posts on this character input problem now, so we're definitely on top of that. Full-screen input is also coming, we're certainly not getting rid of it.

sfrrr - definitely an argument for letting people turn it off, though I agree with ipsi that on-by-default would be best; that would indeed generate a lot of tech support e-mail, particularly from people without enough memory to install the complete set of audio files (and even more from those who don't know they're low on memory).

ssaito - sometimes exiting Pleco completely (using the Quit command in the Dict menu) immediately after unlocking it may help a bit with this, but there's definitely a known bug with settings/unlock codes going missing after a while.

roddy - yeah, it's not likely to make much of a difference in everyday use, and given the number of other problems in 2.0 / features people would like to see added I'd say it should be pretty low on our priority list. It's certainly possible to do tone changes accurately, commercial Chinese text-to-speech systems seem to work well enough, but it would definitely be a lot of work.

 

[daniu]

The neutral tone thing is only going to be a problem when you have a word that is a) outside the 30,000+ ( :shock: ) word set of pre-recorded words and b) has a neutral tone in it. I'm going to guess that is a pretty small set, and not one anyone is going to be using every day. It's not going to affect frequent neutral tone words such as 哥哥, because they will have a specific word-level recording that will take the neutral tone into account.

It wouldn't be that easy to build up tone-correct words from single syllable blocks. You'd need to take into account tone changes - two third tones, bu's and yi's, and also the different pitches of neutral tone. I guess it could be done, but frankly I'm not sure it's worth the effort. It'll be easy to hear when you're getting an ad-hoc built-up word, at that point we'll just need to have a quick look at the pinyin and check what we're hearing is right.

Hi!

Am I right that these cases are actually spoken with a tone by people in Taiwan? I am actually not that advanced that I feel like I really get the problem here ... but I also wonder whether this is also connected to the fact that I learned a big part of chinese sound in Taiwan. Am I right that chinese people from China use the fifth tone for lao3 shi1 de shi1?

regards
Daniel

 

[haraldalbrecht]

That is at least the way my 老师 in 北京 is pronounciating her title: fifth tone on shi.

 

[ipsi]

Pleco won't open PalmDoc files from an SD Card on my Treo 680. Works fine from internal memory, but not from an SD Card. Check my Four Great Novels thread for examples.

Also, when I try adding the flashcards from my NPCR set, it'll happily count up to 955 and then stop. Tried this twice, and both times it stopped on the same number. Damn weird. Going to 'Manage Flashcards' after doing that caused a soft reset. Probably because I removed the battery part way through. Making them smaller didn't help at all. Count still went up to 955 and stopped. What's up with that? Could be a formatting issue?

 

[ipsi]

Found the cause! Mike, you're not dealing with Quote Marks properly: A line like the following:

鸟宿池边树 Niao3 su4 chi2 bian1 shu4 "A bird spends the night on a tree by the side of the pool."

or

元 yuan2 (the same as "块", but used in written language)

*will* cause Pleco to freeze at some point (I think after it finishes reading in the file). I don't know why, but I do know that I was able to import every file without them (about 30-odd) without problem. When I tried importing a file which had even a single ' " ', it froze, and I had to reset. Not sure what's up there. Maybe your database code is treating them as literal, un-escaped quote marks? I'll try replacing them with something else and get back to you. Maybe single quotes will work? Like " ' "...

Ok, it works if I replace " with '. . Yay!

Only problem I have now is that the pinyin is displayed as numbers when I test myself. Why's that? It's a bit annoying... Also, I added them in the form S[T], and it displays it as that when I'd rather it didn't. It's easily fixable though, so I'm not too worried. .

Hmm... Actually, I just noticed that it's cutting off the last character of the Pinyin when it's displaying the flashcard. Buh? That's a bit strange... Yep, definitely being chopped off somewhere in the display as it's still actually stored as the full string. e.g. dou1 is displayed as dou, despite me seeing dou1 when I edit the flashcard.

Also, Pleco crashed when I added a small file with a BOM, but not when adding one without (small being one line). It worked fine when adding large files with a BOM.

Or it did. I haven't tested it since I realised why it wasn't working. But it didn't crash when it encountered a large file with a BOM.

I also have about 40 categories, and it takes a while to load them up, which I'm not too impressed with. It works, but it takes a while. I do have a bit under 2000 flashcards though.

 

[mikelove]

ipsi - yep, I think it's an escape character thing, something we definitely need to do something about; it's not too likely someone would create a malicious flashcard entry consisting of:

ni3hao3"; DROP TABLE cards;

but even just for normal usage all of those characters should be escaped correctly, and injection vulnerabilities are bad coding practice however benign they may be in this case.

I'm not sure why the Pinyin format wouldn't be correctly converted from numbers to tones, they're always stored as numbers in the database so these shouldn't be different from any others. The flashcard DB actually puts simplified and traditional characters in separate fields, and the importer doesn't know that it's supposed to break brackets up and put them in those separate fields yet, so that's why that part of it isn't working. Last character getting dropped sounds like a strlen being needlessly decremented somewhere, should be pretty easy to find...

As for the performance problem, though, is this in Manage Cards or when starting a session or both? Or is it just when you're bringing up the Manage Categories screen? We're actually playing around with changing the way categories are stored (part of the reason Beta 3 is taking so long is that a lot of things are being outright redesigned for it - really want to get it right this time) and one possibility is to actually go back to giving each card a single "primary category" to allow for faster indexing / new sessions; other "secondary categories" could still be used and even searched on but the software would be a lot faster if you started a session based only on primary categories.

 

[ssaito]

I haven't seen this in the bug mentions above: the cursor controls using joystick on my AT&T 8525 no longer work for PD2 (works for PD1, though b/c of wm6 it's not stable).

Also, the history pull down in the entry box (i think that's the little down arrow to the left) doesn't seem to work.

Thanks,
Steve

 

[ipsi]

Yeah. Speaking of not sanitizing database inputs: http://xkcd.com/327/

Anyway, the pinyin is just not being shown correctly. Check 'Manage Flashcards' -> 'Info' to see it. And the string-length bug. Fair enough on the non-splitting of characters. I did see that when I actually tried to edit the flashcards, so yeah. No biggie there.

Manage Cards is pretty quick to come up, but when I ask it for a list of categories, it's really, slow. And it's not even looking for all of them, just the first 8. Well, it's not slow, I guess, but it does take a couple of seconds to list the eight categories it's displaying...

The real performance hit comes when going '[menu]+N' -> 'New Flashcard Session' -> 'Cats', which, for some reason, has gotten distinctly quicker. Weird. It's still slow though, to the point where the whole flashcard screen disappears for a couple of seconds before coming back.

Also, the tones and string lenth problem persist into the flashcard sessions, so I guess it's all located in the same/similar place.

 

[mikelove]

ssaito - hardware button controls barely work at all in any flavor of 2.0 at present, actually, so it's normal that they'd be acting up on your system. And the recent search list hasn't been implemented yet either, the searches are being recorded but we haven't yet written the code to draw the list box that pops up to let you pick one. (ironically enough that's actually the hardest part)

ipsi - one of the best recent XKCDs, though my all-time favorite is still #161. (and #149 is right behind, that was a very good month for him)

So it's just the category list that's slow? But actually generating a list of cards (in Manage or when starting a session) is nice and fast? That's odd... should be easy enough to optimize though. We actually removed some indexing from that table in Beta 2 to speed category creation up a bit, as I recall... is this a new database created with Beta 2 or is there any chance it might be an older one moved over from Beta 1?

 

[ipsi]

Yeah, it's very much the category listing that's slow. Actually generating the list of cards is quite quick. Though I think my flashcard DB got corrupted through lack of use, so I had to delete both flashcard files and reimport. I've also deleted them several times previously (after trying to import lists with an unescaped "), so it's probably not an old file. .

When starting a session, the start session button doesn't always work. It's a bit weird.

 

[mikelove]

Start Session only works from within the Test tab, won't work from the other 4. So that might have something to do with this. Or maybe something else... how did this file corruption manifest itself?

 

[ipsi]

Well, I'm merely assuming that's what it was - I got soft resets on trying both 'New Session' and 'Manage Flashcards', so I deleted the databases, reimported my cards and everything was good.

Or it might have been a reset when I tried to actually start the session. I'm not sure. Either way, deleting helped so I'm assuming corruption.

 

[mikelove]

Hmm... we see a lot of those crashes too, but they pretty much always seem to be cleared up by a soft reset. Could you could send us the flashcard data file (with its associated "journal" file if there is one) the next time this happens?

 

[ipsi]

Sure.

 

[ipsi]

Mike, why can't I attach files in this forum, but can in the others? PM'd what I was going to attach.