Since i moved to new code, the reviews (apart from one word feedback like “crap, shit, worthless”) mentioned that its not acurate. well i now use similarity algorithm which means that user trace is compared against what apps database.
“tree” and “three” have very similar path. Infact half the time even i can’t do it perfectly. Similarly
“ciao” and “chiasso” have a very simiar path.
so I started by first doing a similarity match followed by a sort to reorder but that messes everything up.
“starts” always returns “states”
so what did i do ? well i said “if the word index of 2nd word is greater than 1st word and the difference is greater than 100, show second word before 1st. Thats it. doesn’t do anything elegant.
It solves all the issues i mentioned about but it still make me feel a bit dirty… yikkkkkkeeeeeeessssssss.
I have also tried to patch the runaway blank space bug 🙂 I doubt i’ll ever fully get rid of it.