InformAction Forums

When measuring network speed at http://speedtest.net/ I get significantly different upload speed results when I anonymize requests with ABE:

I invented a general user rule, to anonymize everything except for SELF, causing many information entries appearing on the error console.

[ABE] <ALL> Anonymize from ALL on {POST http://www.speedtest.net/api/api.php <<< http://speedtest.net/ - 12}
USER rule:
Site ALL
Accept from SELF
Accept INCLUSION(SCRIPT, OBJ, SUBDOC) from SELF
Anonymize from ALL
Deny INCLUSION(SCRIPT, OBJ, SUBDOC) from ALL

My measured upload bandwidth goes down from 1.05Mbps to around 0.15-0.19Mbps with this rule.

I have NS 2.2.4 and FX 3.6.24, MZ FX for Ubuntu canonical 1.0. I have not tried any newer, as I have found nothing related in the changelog.

A few comments about how I would like to browse and why I was extremely happy about ABE:
I hate the behaviour of the browsers that they allow playing with non-anonymized inclusion across domains, it is a clear leak to privacy.
I would like this functionality to be opt-in, that means I would like to have a whitelist that allow inclusion from domain1 to domain2 etc to pass my personal data.

ABE user rules allow me to do so, but it seems there is an issue with the performance when anonymizing. (Is it not possible to block looking up the cookies instead of modifying the http request etc,? -- or would it be possible only inside the browser and not in the add-on?)

Best regards,
Jasper

I have found the abe_rules.pdf.

Maybe I only need Sandbox instead of Anon. Does Sandbox strip off cookies as well?

I think it would be worth to have a more fine-grained control than Sandbox and Anon...

I guess I would prefer something like Anon but without removing upload data...
(This allows some scripts to convert cookies etc. into upload data I know)

jasper wrote:My measured upload bandwidth goes down from 1.05Mbps to around 0.15-0.19Mbps with this rule.

Confirmed, on Fx 3.6.25 with NS 2.2.5.

There are going to be major structural changes in NoScript 3.x for the desktop, which was targeted for release by the end of last year, but pesky new web threats kept emerging, especially when HTML5 came out. Etc. Still looking forward to it, and it may address some of your concerns and wishes. Probably not even Giorgio knows exactly what the final product (or release candidate) will look like, or do.

Giorgio is relatively unavailable until the completion of his relocation and connection to his new ISP, but if there isn't a response within a week or two on the issue of the upload performance hit, please post back. I'll be notified by e-mail when you do, and I'll personally ask him to look at this.

Thank you for your patience.