[tor-talk] introducing ahmia search

ahmia ahmia at tormail.org
Mon Oct 1 13:32:59 UTC 2012


Thank you for your reply Mike!

> Am I right in thinking that the way this search engine works is that
> the entire index is sent to the browser initially, and then the
> browser locally searches that index? If so, that's a pretty neat way
> of protecting peoples searches. Obviously, it wont scale, but you 
> know
> that already.

Yes you are correct. Search is done with JavaScript (client side) and 
the entire index is sent to the browser initially. There will be no 
search logs.

> It feels like you're doing the search in the main thread as I type
> into the search field, as it's blocking the user interface a lot. I
> suggest you use javascript web workers for this purpose so you can
> background the searching task onto a new thread. Or just get rid of
> the whole "search as I type" thing. Gimmicks are fine, unless they
> actually make the user experience worse.

JavaScript search is now faster and does not block the user interface. 
This is done now by waiting 1000ms from the very last keyup event before 
laughing the search. Since the web worker cannot access the DOM we 
decided to implement that this way for now. At least it works now 
better.

> Please provide https. At the moment, the search engine can be
> trivially MITM'd and script inserted to log searches. Also, please
> rethink your decision to let Google execute arbitrary code on your 
> pages.

HTTPS it is now (https://ahmia.fi/)! Furthermore, we have removed all 
Google's code and every resource is now downloaded from ahmia.fi - even 
those twitter news.

Thank you very much for your answer! We are always ready to do these 
kind of improvements to our site.

-ahmia, ahmia(at)tormail.org.remove.this-


More information about the tor-talk mailing list