Page 1 of 1

HOSTS file improvement

Posted: Mon Mar 16, 2015 3:35 pm
by gordon451
Like many who regularly use the HOSTS file, it's an important security device as much as a DNS by-pass for important sites.

With this in mind, is there any way the HOSTS file management could include processing of wild-cards?

OK, I know this could be a smoothly paved downhill highway to perdition as we forever strive to get the most comprehensive bloat-ware HOSTS file, but OTOH some of the gynastics we have to do in order to eliminate (in this case) whole ever-increasing families and clans of P(otentially) U(nwanted) S(ites) is mind-breaking and wearisome.

In asking this, I confess I have no knowledge of where or how the HOSTS file is maintained.

Oh, nearly forgot. Some UAC on the HOSTS file? Please? In addition to merely setting permissions to Read-Only?

Gordon.

Re: HOSTS file improvement

Posted: Mon Mar 16, 2015 10:00 pm
by erkinalp
As far as I know, it is possible to implement this. If implemented, it should be implemented in a manner like Unix shadow password file(it appeared after a bug in

Code: Select all

ed
which caused files to be saved in wrong paths and everyone saw password file as the MOTD) . The problem is, you will have to check two places now.

Re: HOSTS file improvement

Posted: Mon Mar 30, 2015 2:09 pm
by gordon451
erkinalp wrote: ...it should be implemented in a manner like Unix shadow password file...
I would not be happy with any solution which involves checking multiple files. The HOSTS file was meant to be a fast local redirect for problematic human-readable host names, which to me seems to be at odds with cumbersome file setups.

I do understand that (in my ignorance) giving wild-card technology to the HOSTS file would involve a major patch to the Kernel. I hope I'm wrong and we are only looking at a "subroutine" if I can use that terminlogy.

Ummm. It's just occurred to me that maybe we need to look at browsers rather than the Kernel? HOSTS files are there to redirect browsers, yes? And all my browsers need to be restarted after modifying it :[ H'mmmm.

Gordon.

Re: HOSTS file improvement

Posted: Mon May 11, 2015 12:29 pm
by gordon451
gordon451 wrote: Ummm. It's just occurred to me that maybe we need to look at browsers rather than the Kernel?
I wuz ronnnng!!!! :D It is in the Kernel! In the DNS client...

So, what's the chances of having a hack to enable the use of wild-cards? OK, this is probably not the easiest hack around, and certainly we need to get ROS airborne before we fiddle with the gubbinses, but can this be done???? :idea: 8-)

Gordon.

Re: HOSTS file improvement

Posted: Thu Aug 18, 2016 6:58 pm
by Oxectirce
gordon451 wrote:
erkinalp wrote: ...it should be implemented in a manner like Unix shadow password file...
I would not be happy with any solution which involves checking multiple files. The HOSTS file was meant to be a fast local redirect for problematic human-readable host names, which to me seems to be at odds with cumbersome file setups.

I do understand that (in my ignorance) giving wild-card technology to the HOSTS file would involve a major patch to the Kernel. I hope I'm wrong and we are only looking at a "subroutine" if I can use that terminlogy.

Ummm. It's just occurred to me that maybe we need to look at browsers rather than the Kernel? HOSTS files are there to redirect browsers, yes? And all my browsers need to be restarted after modifying it :[ H'mmmm.

Gordon.

Based on my testing, a 34 MB hosts file with a few million entries slows down browsing in Windows a bit even when the "DNS Client" service is disabled. So, there is a maximum size beyond which you probably don't want to go.