Pleco for Android 2.4.5 Bug Report Thread

[3x13]

What happens if you simply tap on the paintbrush button again? Does it ignore you? Also, is there a "Done" button in the handwriting input box?

Well, the little arrow in the lower left corner disappears, also the "two fingers" info. So it's all blank on the left then.

 

[mikelove]

I think a list is useful as long as it contains all words which match the actual input field. When the list is (too) long, the user can add a letter to shorten the list. As long as the list is not empty the word he searches can still be in the dictionary, and he can add more letters and shorten the list more - entering more letters when the user still has a chance to find the word is not the problem.

The question is what we're actually displaying a list of - I checked out JED's approach and it seems like for long completions it's not really all that useful; you've got a whole bunch of words in relatively random order, unless you scroll through all of them and check each one by one you're not going to learn whether or not the word you want is there.

Auto-completion produces a list with one randomly chosen word. Looking for a certain word and getting a word that was chosen by any algorithm probably will not help the user. The normal user - without having studied the manual in great detail - believes what he sees and will assume that his word is not part of the dictionary and quit searching.

I still think we could fix that by simply showing the auto-completion in the search field, thus implying that we're guessing what they want but that they can get different results if they keep typing; Google does this and doesn't seem to have a whole lot of people confused. We can certainly consider an option to display a candidate list too, but I'm inclined to make that gray text behavior the default.

Well, the little arrow in the lower left corner disappears, also the "two fingers" info. So it's all blank on the left then.

Any chance that the card itself might also be blank? What happens when you reveal it?

 

[3x13]

Any chance that the card itself might also be blank? What happens when you reveal it?

No, it's a normal card. If I reveal it, I can see it. Also the speaker button works (even if it shouldn't because I switched that off).

 

[mikelove]

No, it's a normal card. If I reveal it, I can see it. Also the speaker button works (even if it shouldn't because I switched that off).

Hmm... did this only happen with this one card or does it happen with every card or does it happen intermittently?

 

[ckatt]

So, the New Cat button in the flash card search results dosn't work correctly. it lets you enter a name for the new category but no category is actually created. i'm not sure if this is something that was supposed to work, maybe is just a hold over from the organize cards screen. But in a few cases i have wanted to add my search results to a new category so if this button functioned it would save me some switching back and forth.

 

[3x13]

Hmm... did this only happen with this one card or does it happen with every card or does it happen intermittently?

Cards are random, so it's a different card each time. And it happens always.

 

[mikelove]

So, the New Cat button in the flash card search results dosn't work correctly. it lets you enter a name for the new category but no category is actually created. i'm not sure if this is something that was supposed to work, maybe is just a hold over from the organize cards screen. But in a few cases i have wanted to add my search results to a new category so if this button functioned it would save me some switching back and forth.

Actually it's supposed to be disabled, since the system doesn't know which parent category that would be created under. Though I suppose we could also consider keeping it and having it add them to Uncategorized.

Cards are random, so it's a different card each time. And it happens always.

OK, we'll investigate further... thanks.

 

[HW60]

Auto-completion produces a list with one randomly chosen word. Looking for a certain word and getting a word that was chosen by any algorithm probably will not help the user. The normal user - without having studied the manual in great detail - believes what he sees and will assume that his word is not part of the dictionary and quit searching.

I still think we could fix that by simply showing the auto-completion in the search field, thus implying that we're guessing what they want but that they can get different results if they keep typing; Google does this and doesn't seem to have a whole lot of people confused. We can certainly consider an option to display a candidate list too, but I'm inclined to make that gray text behavior the default.

I would be happy with a candidate list, even if I had to change the settings to get it.

I looked at Google translator (in case you meant that program). Google translator just tries to find one single translation when you enter a word, and translates what was entered. I tried "run" which was too short with too many results in Pleco, and then chose "run into". In Google you have changing translations with any letter you enter (e.g. "来看，我" for "run i", but normally you get only one result (only with "run" alone you get 12 results alltogether). In Pleco you get changing full-text search results with any letter entered (e.g. "持" for "run i" (because of "to run (i.e. administer)" in CC), but very many results for "run" alone and not necessarily translations as the example shows.

I think a full-text search can also try to produce a translation, but actually the main goal is to find the word in all possible entries, especially also in all example sentences. With "run into" Google gives one result 碰上, with Pleco I found about 60 results in All Chinese Dictionaries. This is quite normal: the user can get a feeling for the use of the word.

I did not realize any sort order looking at the 60 results for "run into" in Pleco, there is no sorting by pronunciation or by dictionary, and it would take some time to find out if some of the results are equal in 2 or more dictionaries, but I do not think that that is a problem: in Pleco I just study the results, and as long as I am interested in more results I can scroll. Otherwise I enter more letters. That is the main difference between translation and full-text search. If I only wanted to know the translation with Pleco, I could use the E-C dictionaries instead of full-text search.

 

[mikelove]

In Pleco you get changing full-text search results with any letter entered (e.g. "持" for "run i" (because of "to run (i.e. administer)" in CC), but very many results for "run" alone and not necessarily translations as the example shows.

It's actually meant to be more like Google search than Google Translate, since that's closer to the kind of search we're doing (think of each dictionary entry as a 'page'). If you type "run i" in a Google search box, you get a list of common queries that match "run i" - at the moment the most popular one of those seems to be a song title - but you only ever get results for one of those queries at a time. Google wouldn't bother giving you results for every phrase that began with "run i" at the same time because they wouldn't be very useful - it would take you more time to sort through them than to select or to finish typing in your search, and the mix of phrases would make it much more difficult for them to make sure the first few results were all relevant (do we list the most popular result for every possible completion first, then the second most popular, third most popular, etc, do we sort all results by popularity even if one completion might be much more popular and therefore dominate the first few pages of results, etc).

I did not realize any sort order looking at the 60 results for "run into" in Pleco, there is no sorting by pronunciation or by dictionary

There's no dictionary sorting because if the same word shows up in multiple dictionaries it's still only listed once - that's the whole idea of our "merged search" thing. Instead of sorting by pronunciation, we sort by a combination of word frequency, placement in the definition (whether it seems likely that this is a clear translation for that word), and presence in more than one dictionary (which of course adds to our confidence in the translation). Not quite as complicated / intelligent as Google's algorithm, but the same basic idea - settle on a single search query and return the most relevant results for that query.

 

[HW60]

I had 2 crashs today ("Leider wurde Pleco beendet" - Unfortunately Pleco was terminated) during flashcard sessions. I will send you the logfile.

 

[HW60]

It's actually meant to be more like Google search than Google Translate, since that's closer to the kind of search we're doing (think of each dictionary entry as a 'page').

I think Google Search and a full-text search have two completely different goals: Google tries to know what the user thinks and wants, and therefore saves all possible information available about the individual user and about everybody. If I searched http://www.pleco.com, the next time I only have to enter "pl" and Google proposes http://www.pleco.com. Even the searches of other people are used to make a suggestion for my wishes.

When I make a full-text search in Pleco, I normally want to know a word or the usage of a word which I do not know. Therefore I am probably looking for the opposite of what everybody is searching for - the relevant and popular words a "Google search in Pleco" proposes I hopefully know already, and if yes, they are not relevant for me. For the search in a dictionary Google translator maybe closer to the needs of the users with their different knowledge (and language btw).

Two examples of my full-text search tests with "run into" which show that full-text search is a great tool which gives me much information to read, though the results are not always expected. Results for "run in":
- from ABC: 到头来 = after all; in the long run; in the end
- from PLC: 乱七八糟 (from the example "the school is being run in a chaotic manner")
I like full-text search and wait for the candidate list!

 

[HW60]

I have again a problem with full-text search: In the main screen (Browse entries ...) at the beginning of Pleco I select language chinese, my user dictionary. I enter "#" - nothing happens - and then "a". Pleco switches to language english and All chinese dictionaries. Now I find lots of entries in ADS, PLC, CC, HDD and OX, and scrolling through the dictionaries I also find entries for ABC, 21C, NWP and DHD, but no entry for my user dictionary. I cannot even select my user dictionary by tap and hold the dictionary button, because the language is english. I have to switch back to chinese, and Pleco switches to All chinese dictionaries and shows entries of my user dictionary; now with tap and hold I can switch to my user dictionary. Entering the next letter "u" the same procedure starts again.
With "#au" entered and All chinese dictionaries selected, selecting language chinese leads to "No Results Found", tapping on the dictionary button makes Pleco switch to PLC and english and results - but no result for my user dictionary.

Most of my dictionaries including the user dictionary are set to "use in dictionary, reader and OCR". The 3 built-in dictionary groups are set to "Full-text only" and "Skip with button".

 

[mikelove]

I have again a problem with full-text search: In the main screen (Browse entries ...) at the beginning of Pleco I select language chinese, my user dictionary. I enter "#" - nothing happens - and then "a". Pleco switches to language english and All chinese dictionaries.

That's expected - the language setting is for the language of your search query, not the language of the dictionaries you're going to search in. (subtle but important difference in this case)

I also find entries for ABC, 21C, NWP and DHD, but no entry for my user dictionary. I cannot even select my user dictionary by tap and hold the dictionary button, because the language is english.

Not sure what's happening with that - any chance you might have disabled full-text searching in your user dictionary? That would explain its not coming up there and only coming up for a Chinese search.

With "#au" entered and All chinese dictionaries selected, selecting language chinese leads to "No Results Found", tapping on the dictionary button makes Pleco switch to PLC and english and results - but no result for my user dictionary.

That's also normal, since "au" isn't a valid Pinyin search but you've forced the software to try to use it as one anyway by tapping on the switch language button.

Most of my dictionaries including the user dictionary are set to "use in dictionary, reader and OCR". The 3 built-in dictionary groups are set to "Full-text only" and "Skip with button".

Those options are the most likely problem here - try turning them off.

 

[HW60]

I also find entries for ABC, 21C, NWP and DHD, but no entry for my user dictionary. I cannot even select my user dictionary by tap and hold the dictionary button, because the language is english.

Not sure what's happening with that - any chance you might have disabled full-text searching in your user dictionary? That would explain its not coming up there and only coming up for a Chinese search.

Yes - I forgot I had loaded a new user dictionary, and I think for new user dictionaries the default value is "Not endabled for full-text search" (i.e. I have to tap on "Enable full-text search") - after that all problems are gone. I had the same problem 2 weeks ago and found the solution by myself ...

Maybe it would be a good idea to make "enabled" the default, even though it is experimental (or set a flag like a red background color for the dictionary name to show it is not enabled).

 

[mikelove]

Maybe it would be a good idea to make "enabled" the default, even though it is experimental (or set a flag like a red background color for the dictionary name to show it is not enabled).

Possibly once the feature is better tested, but right now it's still kind of "experimental" and we'd rather not push it on people too aggressively right away.

 

[sahal]

In card statistics:

When I have one card with a score of 150 and one with a score of 250 in my card set, the score ranges look like this:

101-200: 1
201-400: 1

When I have two cards of score 250 in my set, the score ranges look like this:

101-400: 2

Not really wrong, but not as informative as "200-400: 2" would be.

*****

Also, wouldn't it be more logical to make the "alternating" tests reliably alternate between two repetitions of the same flashcard? Right now they only seem to alternate between each test within the same session, i.e.

Now:
first session: card1 tests for character, therefore card2 tests for pinyin, card3 tests for character etc.
second session: card1 might test for character, card2 would then test for pinyin, card3 for character etc.

What I would find more logical:
first session: card1 tests for character, card2 might test for pinyin etc.
secomd session: because last time card1 tested for character, this time card1 is tests for pinyin, because last time card2 tested for pinyin, this time card2 tests for character etc.

 

[mikelove]

Not really wrong, but not as informative as "200-400: 2" would be.

Yeah, that whole screen is just awful - hasn't been noticeably modified in about 5 years, and it was a crude hack even when we developed it. We've got a long-simmering plan to add graphs and other cool features which we'll hopefully be able to implement in the next year or so, assuming that Windows Phone and BlackBerry 10 manage to avoid getting too popular.

Also, wouldn't it be more logical to make the "alternating" tests reliably alternate between two repetitions of the same flashcard?

That's actually what "Score-based" is for - have you played around with that? "Alternating" to be honest is an option we've been meaning to take out; like a few other features (manual scoring, Frequency-based testing) it's something we've left in for the sake of long-time users but aren't particularly attached to.

 

[TheHunter88]

I think i've found something. Using a Samsung Note 2, when i'm writing 日 (ri4) (as in 日本 Japan) it works great, up until i put the final stroke, then it only shows me 八 (ba1) (as in the number 8 eight).
I'm not sure if it's a bug with the Note, Pleco or something else, i will try and check if it's the same thing with my father's tablet tomorrow.

Edit:
Just noticed that if i add another stroke, no matter what it is, it shows me all the options again with 日 being the first.

 

[HW60]

I think i've found something. Using a Samsung Note 2, when i'm writing 日 (ri4) (as in 日本 Japan) it works great, up until i put the final stroke, then it only shows me 八 (ba1) (as in the number 8 eight).
I'm not sure if it's a bug with the Note, Pleco or something else, i will try and check if it's the same thing with my father's tablet tomorrow.

Edit:
Just noticed that if i add another stroke, no matter what it is, it shows me all the options again with 日 being the first.

For 日 and (sometimes) similar characters, you have to write the first stroke (top down left) and the last stroke (bottom left right) so that they touch in the bottom left corner; if the corner is open, the character 八 is often the only character realized by HWR.

 

[TheHunter88]

It does that regardless off the last stroke. It never did that on the iphone. Also sometimes it will give the correct result, but usually only 八.