[Bloat] Adblock - or another extension - is incorrectly blocking the speed test

[jb]

The warning is correct in that it is probably NOSCRIPT. I think.
All the speed test knows is that an API call to all servers was brutally
failed
in an unexpected way. There is no visibility into what caused the failure,
only
that it should not occur in a clean browser. If you open the console
you can probably see more than the javascript gets told.

For example if the test is given 5 servers to use, it pings them all, and
one
is unreachable, that is fine, no error is generated. The errors accusing
extensions are generated when no other possibilities are left.

I usually see users strike this when they use AdBlock or NOSCRIPT, and
I do believe there is a no-script type extension for Chrome as well. I've
not
pinned down any other cases, perhaps you are the first? but you did
mention noscript so it seems more likely now. I did ask the author
of noscript to ease things by allowing a page to suggest a list of resources
it wants whitelisted in advance, via a line in the html header, and
prompt the user "this page wants to use these resources, allow/deny",
but he said that would be "abused".

Regarding the proxy warning, the test is supposed to be maxing your
last mile link speed and reporting it to the under your ISP. The
majority of proxies seem to be things the user did not even know they
are using like google compression proxy (for chrome on mobile),
or corporate proxies for virus scanning (that perform poorly). If the
proxy is buggy or not transparent enough it breaks the test. For example
if you fetch a large file, then abort when you get 10% of it, does the
proxy fetch the whole file first? does it cache the request aggressively?
If you post, does it suck up the post, then hang without traffic until
it gets the remote response. So a warning that the results may not
be right seemed appropriate. If they are right, great!

Also there is at least one strange proxy+browser app out there, for
accelerating mobile devices. They get huge results and show the
"client" being an ISP IP address, but actually they're doing all the
work in their network. They are probably running a cloud browser
and sending html back to the device, where it is displayed by their
browser, and sending up touch events. A sort of apple screen
mirroring thing.

Do let me know if you work out what it is, perhaps noscript is a little
harder to disable than it might seem?

On Sun, Apr 26, 2015 at 12:15 AM, Jan Ceuleers <jan.ceuleers at gmail.com>
wrote:

> On 25/04/15 13:44, jb wrote:
> > The message appears when the speed test is blocked from reaching a test
> > server.
> >
> > It has turned out to be almost always either Adblock, AdblockPlus or
> > NOSCRIPT
> >
> > The error message is correct, at least, has been so far: it is always an
> > extension
> > and whatever extension it is, does not report to what it is doing, just
> > rips a hole
> > in the browser.
>
> Justin,
>
> I do have other extensions besides Adblock (including NoScript), but
> when I get this error message I had permitted all scripts from sites I
> don't distrust to run (so excluding any google/fb/twitter/... stuff that
> doesn't belong) and as I said I had completely disabled AdBlock.
>
> What I mean about the error message being incorrect is that it doesn't
> tell me what the problem is, and makes an assumption about the cause of
> that problem. So what server does traffic not get through to? I then
> have a shot at finding out which protocol and port so that I can further
> debug.
>
> On another note: why the warnings about use of the speed test through a
> proxy? Why would that invalidate the test? I run a squid instance on my
> side of my DSL line, and I don't notice any significant differences in
> the results with and without proxy.
>
> Thanks, Jan
>
> _______________________________________________
> Bloat mailing list
> Bloat at lists.bufferbloat.net
> https://lists.bufferbloat.net/listinfo/bloat
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.bufferbloat.net/pipermail/bloat/attachments/20150426/6550f730/attachment-0002.html>