Timothy Guan-tin Chien

But I am not interested in creating my own byte to float32 conversion...

Let's focus on the string part of the data first. Here are some ideas after reading some article about punycode and delta compression: Around 50% of the characters following the...

> Around 50% of the characters following the first character has a delta smaller than 0xff, so if we store the delta rather than the actual USC-2 code, we could...

#35 is an invalid issue. We could still implement Pinyin with Encoded sounds as the backing sound encoding. One thing that need significant change would be the partial sound matching....

http://stackoverflow.com/a/33266296/519617 sw-precache will work too but seems to be an overkill...

https://developer.mozilla.org/en-US/docs/Mozilla/Tech/Places/Frecency_algorithm

Sorry for the late reply. This is probably a bug in the browser (not being able to draw a word with the size of `11445`!). I did see an error...

Yes, the number is the weight but unless your weight factor is way smaller than 1 I assume 11445 means font-size > 11445px? DOM rendering is quiet different than 2d...

The reason why there is a 4px limit is that smaller grid sizes make the thing run slow. I probably make sense to remove it though, we can always add...

Hi! > In my case, the weight of each words is a number of occurrences. I use the weightFactor function in order to normalize it. The problem is depending on...