837715 - Reduce compression rate for js source

Status: RESOLVED FIXED

(Core :: JavaScript Engine, defect)

Description

[Jeff Muizelaar [:jrmuizel]]

Attachment: Reduce compression rate for js source

This cuts the time to compress gaia-email-opt.js from 0.63 seconds to 0.33 seconds.
The result should still be smaller (391K from 321K before) than the compressing with snappy or lz4 (528K), but decompression time will be worse. Fortunately, this will only penalize Function.toSource() which I think is an ok trade off.

Comment 1

[:Benjamin Peterson]

Yes, definitely; penalize toSource() all you want. The real concern is memory usage. I'm CCing njn to see what he thinks about this. Could you possible see what the difference is memory-wise on desktop? There's a "script-sources" entry in about:memory.

Comment 2

[Nicholas Nethercote [inactive]]

"script-sources" numbers would be good.  Having said that, B2G has tighter constraints than desktop regarding both time and memory, so I'm ok with whatever works best with B2G.

Comment 3

[Jeff Muizelaar [:jrmuizel]]

(In reply to :Benjamin Peterson from comment #1)
> Yes, definitely; penalize toSource() all you want. The real concern is
> memory usage. I'm CCing njn to see what he thinks about this. Could you
> possible see what the difference is memory-wise on desktop? There's a
> "script-sources" entry in about:memory.

I'd expect a similar difference: about 15%-20% worse. The difference for minified source will be less, the difference for unminified larger.

gaia-email-opt.js - 22% increase
jquery.min.js - 16% increase
ga.js - 11% increase
gmail.js - 14% increase

Comment 4

[Jeff Muizelaar [:jrmuizel]]

This should give a significant improvement in the email app start time.

Comment 5

[[:fabrice] Fabrice Desré]

Please land asap!

Comment 6

[Jeff Muizelaar [:jrmuizel]]

https://hg.mozilla.org/integration/mozilla-inbound/rev/eb8f60c782da

Comment 7

[Ryan VanderMeulen [:RyanVM]]

https://hg.mozilla.org/mozilla-central/rev/eb8f60c782da

Comment 8

[Ryan VanderMeulen [:RyanVM]]

https://hg.mozilla.org/releases/mozilla-b2g18/rev/2f50bc4f27ed
https://hg.mozilla.org/releases/mozilla-b2g18_v1_0_0/rev/6f06092e3628

Comment 9

[Jeff Muizelaar [:jrmuizel]]

Quite surprisingly, this seems to have eliminated the kraken win that happened in bug 836515. I can't think of any reason that would happen as I would've expected an improvement.

Comment 10

[:Benjamin Peterson]

I dont' see any regression on http://arewefastyet.com/.