Url filters - The Discipline of Annoyance & Demise
Saturday, 17. February 2007, 01:28:28
Note: title kindly adapted from Emperor's 2001 album
Lets do a little experience.
1. Go to the website http://www.asdf.com/
2. Select block content from the right-click menu
3. Add http://www.asdf.com/* to the blocked urls.
4. Now open this url. Here's what you see
5. Go to a unexistent website like http://foo.bar/. Here's what you see
Can you make the difference ? No ? Look a little closer, approach the screen a bit, observe from different angles. Still can't see the different ? Well, the second website has some red in it.
What does this mean ?
Whops ! You stumbled upon a invalid domain and Opera cannot connect. But what about the 1st case ? Opera deliberatly blocked the url.
Now imagine that you just downloaded one of those ad lists online. You start to wonder..hum ! Opera can't render this webpage. It's a Google conspiracy for sure.
Most filter files encoutered online are too generic, and hit too many false positives.
Consider the case described in this thread.
The user complained because Opera blocked an image which add "/ad4" in it's base64 content. Well too bad. You downloaded a too generic filter file, now you hit some false positives. But he was lucky to solve the problem with such a simple case. If you check the OP's filter file, whenever there should be either a dot and an wildcard, or wildcard and dot, there is no dot, becaise the person that did that filter file didn't consider that ad4 could ever be part of a valid url, which that rule would happily block.
Now, why did I initially approach this subject with the asdf.com example ?
Simple. Opera currently does not provide any way to tell the user that some content is blocked. Iframes get blank, not loaded scripts go to the error console and that's it. Opera definetly needs a friendly error page telling the user that a url was blocked. If so, the user would then inspect either the info panel, or whatever, to realize that there is blocked content.
As an extra case, I once saw a user complaining he couldn't open any website that started with http://count*
Where do I recognize that ? Most statistic website's domains start with count. So the poor used downloaded a bad filter file, and actively contributed for Opera not being part of statistics online. The author of that filter file must be proud.
Conclusion: don't accept generic filter files out of ignorance. First inspect them. Tamil has a almost good list, but I consider it too generic. My personal filter file has a very little amount of generic rules. The rest is all full domains, and I rarely see an ad.
Happy browsing (without annoying ads).
Lets do a little experience.
1. Go to the website http://www.asdf.com/
2. Select block content from the right-click menu
3. Add http://www.asdf.com/* to the blocked urls.
4. Now open this url. Here's what you see
5. Go to a unexistent website like http://foo.bar/. Here's what you see
Can you make the difference ? No ? Look a little closer, approach the screen a bit, observe from different angles. Still can't see the different ? Well, the second website has some red in it.
What does this mean ?
Whops ! You stumbled upon a invalid domain and Opera cannot connect. But what about the 1st case ? Opera deliberatly blocked the url.
Now imagine that you just downloaded one of those ad lists online. You start to wonder..hum ! Opera can't render this webpage. It's a Google conspiracy for sure.
Most filter files encoutered online are too generic, and hit too many false positives.
Consider the case described in this thread.
The user complained because Opera blocked an image which add "/ad4" in it's base64 content. Well too bad. You downloaded a too generic filter file, now you hit some false positives. But he was lucky to solve the problem with such a simple case. If you check the OP's filter file, whenever there should be either a dot and an wildcard, or wildcard and dot, there is no dot, becaise the person that did that filter file didn't consider that ad4 could ever be part of a valid url, which that rule would happily block.
Now, why did I initially approach this subject with the asdf.com example ?
Simple. Opera currently does not provide any way to tell the user that some content is blocked. Iframes get blank, not loaded scripts go to the error console and that's it. Opera definetly needs a friendly error page telling the user that a url was blocked. If so, the user would then inspect either the info panel, or whatever, to realize that there is blocked content.
As an extra case, I once saw a user complaining he couldn't open any website that started with http://count*
Where do I recognize that ? Most statistic website's domains start with count. So the poor used downloaded a bad filter file, and actively contributed for Opera not being part of statistics online. The author of that filter file must be proud.
Conclusion: don't accept generic filter files out of ignorance. First inspect them. Tamil has a almost good list, but I consider it too generic. My personal filter file has a very little amount of generic rules. The rest is all full domains, and I rarely see an ad.
Happy browsing (without annoying ads).
WIDGETIZE