noscript slows down smugmug

[guest]

I'm experiencing massive slowdowns using my Smugmug website and NoScript. With Noscript enabled, a page takes 20x longer to load than without Noscript. Running MacOSX10.11.6 with NOScript 10.1.8.8 on Firefox 61.0.1. All scripts are "trusted" on Smugmug. While the slowdown time varies some, it is always much much slower with NOScript enabled.

[Giorgio Maone]

Could you point us to the exact URL (if it's public)?
Does the problem persist with latest development build (10.1.8.9rc4)?
Thanks!

[therube]

https://realrail.smugmug.com/Trains

Don't know that I'm noticing any difference?

[guest]

rez.photography. Go to any gallery, then click on a particular one. For example, https://www.rez.photography/Special-Pro ... teo-Coast/

[guest]

Yes, it happens with the rc version also. Here's some numbers I got using the loadtime plug-in
With NoScript (RC)

Loaded 2018-07-25 09:07:55.281
Load Time 38.24 s
Redirect n/a
Domain Lookup 0 ms
Connect 0 ms
Wait for Response 221 ms
Response 761 ms
DOM Processing 37.94 s
Parse 1.16 s
Execute Scripts after Parsing 1 ms
DOMContentLoaded Event 43 ms
Wait for Sub Resources 36.74 s
Load Event 3 ms

Cleared cache, disabled NOScript, then hit reload

Loaded 2018-07-25 09:11:59.453
Load Time 3.23 s
Redirect n/a
Domain Lookup 26 ms
Connect 210 ms
Wait for Response 190 ms
Response 888 ms
DOM Processing 2.77 s
Parse 1.06 s
Execute Scripts after Parsing 0 ms
DOMContentLoaded Event 35 ms
Wait for Sub Resources 1.68 s
Load Event 2 ms

[therube]

NoScript 10.1.8.9rc6

Code: Select all

NoScript
100 requests, 7.82 MB / 4.72 MB transferred, Finish 4.23 s, DOMContentLoaded 1.10 s

Safe Mode
110 requests, 8.04 MB / 4.79 MB transferred, Finish 5.98 s, DOMContentLoaded 1.29 s, Load 4.70 s

I actually wasn't expecting that, that it would be so close.
Am actually surprised that it is.

Safe Mode has that Load figure that was not given with NoScript enabled.
Though... that looks like it is going to be because of, https://forums.informaction.com/viewtop ... 929#p97929.

To me, those numbers are within tolerance, even the 1 s Finish time...
Though... I guess it is true.

Now, can't say what those numbers particularly mean?
Finish? Just what is finished?
And how does Finish differ from "Load"?
Load is not determinable, presently - with NoScript enabled.

So excluding Load, I'll say with or without NoScript, page loading time is within tolerances.
Depending on what "Load" means, & once we're able to see its' figure (in a meaningful way) with NoScript enabled, for now, I've got to say, close enough.

Seat of the pants, gives me the impression that Safe Mode is "cleaner".
Though the numbers I've put together, so far, say otherwise.


Cleared; Cookies, Cache, & Offline Website Data, between runs, &/or restarted FF between timings (from within Browser Console -> Network).
Using that methodology, for better or worse, so maybe the method if flawed.


Looking at "performance".
Safe Mode on the left, NoScript on the right.



I would clear cache..., then force reload the page, a few times in succession.
With NoScript enabled, I'd say higher CPU & longer duration.

So all in all, close enough, but with conflicting numbers or counter to expected.

?

Looking at "performance".

Ah, you're going to have to take that with a grain...
The screenshot, the CPU, Mem, I/O, is only going to be of the parent firefox.exe process.
I initially thought it was showing, I assumed it was showing an accumulation of the entire "multi-process" firefox process, but it is not.

Though between Safe Mode & starting Normally, FF does use more RAM.
In my Profile, as it is, (& with NoScript, & now Load Time, being my only extensions), FF ram changes from ~220 MB to ~330 MB.
(Now, extensions are not the only things that differ when you run in Safe Mode.)

[therube]

Oh, Load Time 0.2.
I don't know that that is going to be accurate due to the bug in NoScript (mentioned above)?
Also, seems you must make sure you clear Cache... (at the least) between runs, I would think.

[Giorgio Maone]

It was actually a quite difficult thing to spot: a 0 bytes "cookiemonster.dm" subdocumnent load (a tracker? However neither Firefox's built-in tracker nor uBlock seemed to agree) being stalled by a responseFilter because, being empty, no data triggered the ondata() event. In practice, the page loaded just as fast, but the feedback (Firefox's status message, throbber, and performance measurements) was messed up by this missing "load completed" signal.

Should be fixed in latest development build, thanks.
v 10.1.8.9rc7
=============================================================
x Fixed some resource loading feedback glitches
x [XSS] Updated HTML event attributes matching
x Updated TLDs

[Guest]

10.1.8.9rc7 works for me. Thanks for figuring this one out.