Page 1 of 1
noscript speed regression
Posted: Sun Aug 07, 2011 4:58 am
by ifdj
For the past week with the latest Firefox Nightly and noscript dev versions the browser lags, which I've never seen before. Minimizing or un-minimizing a window causes a slight delay that wasn't apparently before.
Another example is the sunspider test, with noscript disabled i get 250ms but with it enabled i get 550ms.
OS is linux, only other addon is adblock plus
thanks
Re: noscript speed regression
Posted: Sun Aug 07, 2011 4:26 pm
by therube
If you update the the latest #dev,
v 2.1.2.6rc6, does it improve (or are you using rc6 already)?
Re: noscript speed regression
Posted: Sun Aug 07, 2011 8:39 pm
by ifdj
yup, was running that version of noscript when i made the post yesterday, tried it again today with noscript 2.1.2.6rc6 and the Aug 7th nightly with the same results.
If theres any other info you need please let me know
Re: noscript speed regression
Posted: Tue Aug 09, 2011 3:38 pm
by Giorgio Maone
Confirmed, that looks like a Nightly-only problem, and not depending by any recent change in NoScript.
Could you try to find a regression window?
Re: noscript speed regression
Posted: Tue Aug 09, 2011 4:47 pm
by Giorgio Maone
OK, I found the following:
- The problem started with the July 29th nightly build
- The problem is not NoScript-specific, but it affects any JavaScript-intensive page where window.watch() is used
- You can work-around by disabling the "ab" surrogate, which uses window.watch on all HTTP pages: just set the noscript.surrogate.ab.sources about:config preference to an empty string
Filing a bug report at
https://bugzilla.mozilla.org, since this is definitely a Firefox bug.
Re: noscript speed regression
Posted: Tue Aug 09, 2011 8:26 pm
by Giorgio Maone
Re: noscript speed regression
Posted: Fri Aug 12, 2011 5:02 am
by al_9x
Is the ab surrogate intended for
http://www.antiblock.org/ ? If so, it's obsolete.
Currently their script relies on XMLHttpRequest.open throwing an exception for non-allowed domains ("Component returned failure code: ...") And their code and state is protected in a closure.
Since this is a general countermeasure to external script blocking, it probably makes sense to do something about it? I guess you could patch XMLHttpRequest.open or perhaps you could get it to not to throw without patching?
Re: noscript speed regression
Posted: Fri Aug 19, 2011 10:51 pm
by Giorgio Maone
al_9x wrote:I guess you could patch XMLHttpRequest.open or perhaps you could get it to not to throw without patching?
The way to make it not throw is setting
noscript.forbidXHR to 0.
Otherwise a surrogate should wrap open and swallow the exception.