Today's update brings improvements to the Dictionary, both on Study and Review pages.

• Tweaked visuals (looks a bit nicer, screenshot)
• The kanji reading is now highlighted in all the results of the dictionary! (except special readings such as "たばこ" 煙草 -- in which case there is no highlighted portion)
• Increased font size of the kanji & kana in the Dictionary list for legibility
• Switched highlighted reading style back to red underlined text (which I think is less distracting when reading the entire word, also may help color blind users)
• Performance improvement: implemented a cache for 3030+ RTK characters, which should improve response times (though in practice the main source of lag is server-client response, so it can vary a lot -- I typically see 600ms to 1sec from Belgium, the servers are in US)
• Increased the limit of results to 50 words (it was 30 before -- note the results are always sorted by priority so typically you'll find the word you're looking for)

What does "No common words" mean?

Koohii's dictionary focuses on the 16,000+ "priority" entries as defined by JMDICT (see Dictionary section of the documentation). This is the same source then eg. jisho.org. In fact 670+ kanji out of the 3039 kanji combined between Old & New RTK editions, do NOT have a "priority" entry. In this case the Dictionary says "There are no common words (...)". For example, if you lookup "prosperous" on jisho.org notice none of the words are tagged "common" or "jlpt". As you can see these characters are often used in unique place names. These charcters are of course important in RTK as "building blocks" for more complex characters.

Some statistics

Out of ~3030 RTK kanji (Volume 1 & 3):

• 479 with only 1 "common" word ("common" means a priority entry in JMDICT)
• 257 with 2 "common" words
• 193 with 3 "common" words
• 1870 with 1 to 5 "common" words
• 250 have more than 30 "common" words
• 120 have 50+ words (in which case the top 50 words by priority are shown)

Known limitations:

Dictionary entries for non-RTK kanji do not have the highlighted readings as they are not cached. This is a performance consideration. I could cache more data, but 99.99% of the time users will browse RTK kanji so it's not really worth the hassle and extra storage.

This update closes Github issue #122..