On the phone, I never see the search history as on the comp, that's a pity, because I do repeat searches. Except a few days ago, I briefly saw the search history and thought this bug was fixed, but no, it was just one time. Another small thing is that it took me months to realize I could filter words and kanjis, because, well, I don't associate a cog button to filtering, I thought it would be options about the display of answers, or just didn't pay attention. Maybe it's just me, but a funnel button would have told me from the start that I didn't need to switch to Aedict whenever I wanted to filter out. And yes, a funnel in itself doesn't actually filter stuff (you have to imagine there's some kind of filter at the entrance of the funnel), but over 20 years ago I guess Excel created this association, and it's very widespread now, no? Anyway, a last detail is that now that the word dictionary indicates frequencies, it woul be nice to have a line for frequency filters, with maybe 3 or 4 options, like top 2000, top 5000, top 10000 and "common" (that's 21000 words, right?)... Or something like this? And if there are frequency indications for the kanji in the database, maybe equivalent filters in the kanji dic. For example "top 1000" and "top 2000" and maybe 2 more?
And a couple more things. The way to perfection is asymptotic... On this screenshot I'm trying to add the translation to my private sentence with 下降, but I can't see what I type. Speaking of which, why not let one add the translation immediately (together with the Japanese sentence), instead of having to go back to the word, select the sentence and edit it? Also, is there any plan to allow users to see sentences of experienced users, for words with no "official" sentence or only one or 2, with a warning the sentence might be wrong? When I need a sentence, I pick one on Weblio, I don't think the quality is worse than Tatoeba. I'm sure the private sentences of experienced users are at least as good. In maybe 50 or 100 words, the beginning of the French def is missing, something was probably wrong with the format. During a test, after I answer, I can edit the def, copy and paste the real def from the Tenjin app (which handles correctly the non English defs, it has more Russian defs too), and it's fixed. But from the dictionary I can't do that. Couldn't the "report mistake" button be replaced with an "edit" button when a non English dic is active? With the current system, I'll have the possibility to fix 履く only when it will appear in a quiz. And of course it's unlikely that I'll remember that this word also mean "to affix a bowstring to a bow", so most likely I'll get the answer wrong. Sure, there's the "mark as correct" button, and for such an obvious word that will be OK, but the many other cases are not so obvious. And it would just make more sense to fix mistakes when I see them, no?
Last thing, the kanji dic shows neither frequency indication of the displayed kanji, nor of the words using it.
I think I can fix/add everything in your first post. The recent searches don't appear because there is really no space for them and the keyboard on the majority of mobile devices, and I did not want a super short recents box.
As to the second:
Let me fiddle with the position of the box appearing. You can add English like this: JAPANESE=ENGLISH
I thought it already said that, but maybe not.
Can you give me some examples of the definitions being cut off? If it is an import issue on my end, then I'd rather fix it that way.
At some point, I will make it possible allow viewing of data added by others (sentences, custom definitions, etc.). Still thinking on how I want to do it.
Kanji freq is not displayed, no. I do not plan on adding the freqs to the words listed at the bottom, at least not right now.
Thank you very much. Ah, the absence of the recent search list makes sense, then... But it's strange because I was pretty sure I saw this list, once... I'll take 1 screenshot if it happens again. How about displaying the last 5 (or whatever fits), with the possibility to scroll down to the 20th? The edit box is higher and I could add the translation of my sentence, thank you. I can't remember other examples, sorry. They're common words, JLPT N3 N2, mostly, for which only the most unusual meanings were there, just as with 履く above. If you can see which defs I modified, and specifically, that I made significantly longer (say, at least +12 characters), then you'll have the list of the problematic defs that I caught. The other custom defs are usually a simple rearranging of the translations. In each case of "normal" def missing, I added it from Tenjin and hurried to forget the super-unusual meaning. If the problem is not at import, there could be a second possibility: There are 2 sets of non English defs, and the one you used is worse than the other for French and Russian (I could check Spanish too, and go through N4 N5 in German to compare). However, I assumed that the problem is at import, because the unusual defs are the same (each time I paid attention) as the unusual (2nd, 3rd, 4th...) defs given by Tenjin.
Checking the N1 list, I quickly found 3 examples of words with no French def in Renshuu that have a French def in Tenjin: 改悪, 海運 (bad def, but present) and 改修. That could help clarify the origin of the difference between the 2 dictionaries. If more examples are needed, I think any JLPT word has a French def in Tenjin, so it's likely any JLPT word with no French def has a problem. (although there is only one def for each word, so alternate spellings are tricky). OK, there's clearly another set of non-English defs, that is better. I just realized that Tenjin offers defs in 51 languages. Quite a few have around 40,000 defs, apparently, German has 100k. (I should totally practice N5 in Greek and N4 in Portuguese to make those reviews interesting :) )
Let me know if you find out where Tenjin sources its definitions. I miiiight be able to consider this for the next version of the site!
Besides JMDict, it uses Princeton's Openwordnet http://compling.hss.ntu.edu.sg... Apparently each meaning of each word in English is connected to words in other languages, so it's a giant set of meanings instead of words. It uses Wiktionary, too. I don't know exactly how, but somehow it doesn't give absurd result when you make a Ja>Fr query, you don't get the kind of absurdities that Google Translate still offers due to the fact that all Google pairs are with English and that therefore other pairs are translations of translations, and Google still translate words instead of meanings. You can make queries here to see how the results looks: http://compling.hss.ntu.edu.sg... You can click on any meaning reference number and get from 7 to over 30 languages. This page however doesn't show the whole thing : no results are shown for German and Russian, which have both decent results in Tenjin. So the whole thing is better than seen in this query page. This also shows Tenjin's import wasn't flawless, and that therefore you can build the best available multilingual dic for Japanese. If I check 海運, Tenjin gives me only radionavigation, which is very specific, whereas Princeton says "navigation, radionavigation". The actual translation would probably be "navigation maritime", I guess (yes, we put the words in the right order, that's much less confusing!), because "海運" kinda seems to exclude river and lakes (or not?), but hey, if you get "navigation" in there, that would be great. Also, you can see if you want to merge it with JMDict for English. Still for 海運, it would add "ship traffic" to "maritime, maritime transportation"