Use webworker for bzip2 decoding about kiwix-js HOT 2 CLOSED

It will not be that simple.
My first approach was to use a webworker to call the whole localArchive.readArticle(title, callbackFunction);

But Web Workers can not access the variables of the caller. You have to pass the values through messages.
A File object is not clonable, and the dump files are too big to be cloned in memory.

That's why my first reflex was to use Web Workers only when DeviceStorage is available.
Anyway, except if my tests were wrong, a WebWorker does not seem to have access to navigator.getDeviceStorage() function.

So it seems this approach will not work at all.

Maybe it would work to use a Web Worker that encapsulates only the bzip2 algorithm.
But, in this case, both the input byte array and the output string will have to be cloned. So there will be an overhead.
And the recursive algorithm that reads an article will have to be modified to handle the fact that a Web Worker output is asynchronous

from kiwix-js.

I managed to implement it on the bzip2 algorithm itself.
It works pretty well : the spinner icon is turning while decoding the article, and the User Interface is not blocked.

I measured the overhead, based on the unit test "readArticle" on the small dump : it is 14% longer with the Web Worker (395 ms against 341 ms) on my computer.

Even with this overhead, it appears worth it to me.

from kiwix-js.