[HTTPS-Everywhere] Speeding up Firefox start with SQLite

[Jacob Hoffman-Andrews]

On 01/12/2014 04:34 PM, Claudio Moretti wrote:
> how does the schema handle duplicate targets?
The schema works fine with duplicate targets, but as you point out, the 
JS code I wrote ignores the possibility of multiple results. The 
validator complains about duplicate entries, so I wasn't sure what 
behavior to implement here. I've changed it to include the duplicate 
entries.

On 01/12/2014 01:23 PM, Vijay P wrote:
> You could start a webworker to copy everything from the DB to RAM in 
> the background when you start up...
I didn't have time to learn about webworkers and add the code. Instead I 
added a compromise: At startup we load just the list of available 
targets, so we only have to hit the DB when we know there is a target 
match. This speeds up browsing significantly and only costs ~300 ms at 
startup on my 6-year old laptop. This also helps keep the initial memory 
footprint down (411k vs 3.4M), and most of the rulesets never need to be 
parsed since they are never visited.

If you're interested in writing a webwork to eliminate that remaining 
300ms delay, I would be interested in that!

Also, it appears that HTTPSRules.init gets called three times during 
startup, which amplifies the slowness: 
https://github.com/EFForg/https-everywhere/issues/84.

New version at 
https://jacob.hoffman-andrews.com/hacks/https-everywhere-jsha-sqlite-demo-v2.xpi 
(.asc for sig).