Upgrading Firefox does disable extensions. That will continue to happen past 1.0 -- extensions for Firefox 1.0 will not automatically work in Firefox 1.5. That is the price for using third-party addons, not for using "beta software".
That said, Firefox 1.0PR+ does try to find compatible versions of your extensions when you upgrade. If the extensions you use are well-maintained, you won't lose extensions when you upgrade from 0.9+ to 1.0PR+.
I found two of those holes. I did not find them by looking at the source code. So you're wrong:)
Of the 62 security holes I have found in Mozilla and Firefox, I only found four (217195,162409,249332,87980) by looking at the source. Even then, I didn't find all the holes by reading through large amounts of source. I found 87980 by investigating an error I saw in the JS console during normal use.
I found 162409 during John Keiser's presentation about a feature he maintained. His slide said something like "Session history uses keys to recognize form controls when you return to a page: Tagname>InputName>InputType>FormName>IndexI nForm". I raised my hand and asked "Are those actual greater-than characters in the keys?". He said yes. I asked "Isn't that a security hole?" He said he didn't think it was, because then there would just be too many greater-thans. After the lecture, we worked together on an exploit, and then he fixed the bug.
I'd probably find more code-level holes if I spent more time looking at source code:)
Only five of the ten security holes involved memory management (buffer overflows, etc). Two were file permission problems and three were logic errors. So avoiding manual memory management would only have kept out half of the holes.
Most of Mozilla uses safer abstractions such as reference-counted pointers and string classes (which are, of course, implemented using manual memory management). But Mozilla manages memory manually in a few places where speed matters a lot, such as image rendering. I think that's the right strategy, even though it occasionally introduces security holes, because overall speed relative to other browsers is important.
The delay isn't an insulting, newbie-proofing feature to "make sure you read the dialog". It's a fix for a real security hole: a race condition involving human reaction time. Because it is a fix for a real security hole, there should not be a setting in about:config to reverse it. I'm disturbed that I haven't seen fixes for the same security hole in Opera and Internet Explorer.
If you know that your worst-case reaction time is less than two seconds, you can change the delay.
I listed it as a "major bug fix" in the changelog I created. That does not mean that it was treated as a "major bug" before it was fixed, only that I thought it was a notable fix (at least for the kinds of people who would read a changelog).
Maybe the post meant "0.9 must be ported, or I will mark Firefox (0.8) as broken due to security issues". I'm not familiar with the process and jargon surrounding distributions, so I can't be sure.
Look up the "halting problem" sometime. It's impossible to algorithmically prove that a computer program is correct (bug-free, etc.).
It is entirely possible to write proofs that programs are correct. In the Advanced Algorithms class I took, many of the homework problems were proving that short pseudocode programs were correct. (Proofs can be verified algorithmically, but that's not important the debate over whether it possible for software to not have security holes.)
You misunderstand the halting problem. What is impossible taking an arbitrary program as input and always being able to say either "This program halts" or "This program does not halt". But if you're also allowed to say "I don't know" for some input programs, it is no longer impossible.
Here's a dumb example:
if (program == "")
return HALTS; else if (program == "while(1);")
return DOES_NOT_HALT; else
return DONT_KNOW;
Btw, there's something called "proof-carrying code", but I don't know much about it.
Firefox uses "I'm Feeling Lucky", which always goes straight to the first hit. IE with the Google Toolbar uses "Browse By Name", which only goes to the first hit when Google is sure the first hit is the correct one. Otherwise, it displays search results.
As far as I know, short of a dumb terminal, there's no system in the world that can prevent users from doing dumb things to their files. It's the ones that screw up the system that need to be prevented.
As you said, an operating system can't prevent users from doing dumb things to their files, but generally prevent users from doing dumb things to system files. On the other hand, a web browser should prevent web pages from doing things to users' files. When it doesn't, we call it a security hole in the web browser. On a system with only one user, this kind of hole is more severe than an operating system hole that lets the malicious code then gain root privileges and muck with system files.
I believe not serving the user would be an exception, rather than a rule.
Sure. But that's true of commercial software as well.
Sure, they might not prefetch pages, but if they did don't you think a lot of websites might start to simply block Mozilla, because it sucks up all their bandwidth?
Yes, I think that would happen.
What are you talking about it will not display the alt attribute without a title attribute? I assume that would be part of the rendering engine and this bug would be shared with Firefox? This doesn't occur in Firefox.
Go to www.alltheweb.com and hover over the logo. You'll get a tooltip in IE but not in Firefox, because the text is in an alt attribute. On alltheweb.com, it makes sense for there not to be a tooltip, but there are some sites that have crucial information in tooltips and incorrectly use the alt attribute instead of the title attribute to create the tooltip. The alt attribute is meant to replace the image when it is not shown (e.g. for blind users, lynx users, and users who have disabled images). Alternate text tends to make poor tooltip text and vice versa. Read bugs 25537,74241,41924 (or just Hixie's comments in those insanely long bugs) if you're interested.
Also, I don't know where you got your browser, but my copy of Firefox's help menu has "Release Notes" and "About Mozilla Firefox".
I'm using a branch nightly build and I see those items. I expect them to be there in Firefox 1.0 Preview Release.
While I can train a canonical "Mom" not to answer emails that say "Your bank needs to verifyt your mother's maiden name and social", I'm not so sure I can protect her against these other than to say "NEVER, EVER LOOK AT ANY EMAIL FROM A BANK OR EBAY". If you're running an auction, just keep going back to eBay for the info.
What's so hard about "Always check that you're actually on ebay.com after following a link from an e-mail"?
It seems to me that, to be 100% sure, you must audit the source code line by line.
Given how long it takes most security holes in popular open source software to be found, I think we can safely say that nobody competent enough to audit the source of software they use does so.
And when it comes to that, Free Software is highly desirable, because it always tries to serve the user, above all other considerations.
Some examples where that is not true:
Mozilla does not display the alt attribute of an image as a tooltip when there is no title attribute. This serves blind users of all web browsers by encouraging web authors to use the alt and title attributes correctly, but it annoys some Mozilla users.
Mozilla does not pre-fetch every link on pages visit. While this would benefit users (at least those who don't pay for bandwidth), it would have a negative impact on web sites.
Mozilla supports autocomplete="off" (disables password manager for a specific form) and Cache-control: no-store (disables cache for a specific page), primarily due to blackmail from banks that would block Mozilla otherwise.
Firefox's Help menu contains "Tell a Friend" and "Promote Firefox" menu items. These support the Mozilla Foundation's marketing goals; they are not there to benefit users.
Firefox will not import your home page from Internet Explorer.
Re:Where do you draw the line?
on
The Spyware Inferno
·
· Score: 2, Interesting
AIM is advertising-supported software because it displays its ads in the AIM window.
Kazaa is adware because it displays its ads while you use Internet Explorer. Pop-up adware often makes it difficult for users to tell what application the ads sponsor, which IMO is the point at which it becomes evil. I don't know whether Kazaa's ads say "This ad is shown using Claria technology to sponsor your use of Kazaa. To stop seeing these ads, uninstall Kazaa". I'm not going to install Kazaa to find out.
Hmm, I didn't realize "Open Source" had to be capitalized.
Re:Biologically speaking, how...
on
RGB to become RGBCMY
·
· Score: 2, Informative
If one had the technology to vary the intensity of red, green, and blue over an infinite set of real values, then RGB would be able to perfectly replicate any color.
Wrong. Take a look at a CIE Chromaticity diagram and you'll see that no matter what three wavelengths you choose as your primary set, there will be some colors you can't mimic.
Slashdot Mirror
What is your recommendation? What do I *do*?
Give up. Read and post to Slashdot while at work.
Upgrading Firefox has never wiped my preferences.
Upgrading Firefox does disable extensions. That will continue to happen past 1.0 -- extensions for Firefox 1.0 will not automatically work in Firefox 1.5. That is the price for using third-party addons, not for using "beta software".
That said, Firefox 1.0PR+ does try to find compatible versions of your extensions when you upgrade. If the extensions you use are well-maintained, you won't lose extensions when you upgrade from 0.9+ to 1.0PR+.
I found two of those holes. I did not find them by looking at the source code. So you're wrong :)
I nForm". I raised my hand and asked "Are those actual greater-than characters in the keys?". He said yes. I asked "Isn't that a security hole?" He said he didn't think it was, because then there would just be too many greater-thans. After the lecture, we worked together on an exploit, and then he fixed the bug.
:)
Of the 62 security holes I have found in Mozilla and Firefox, I only found four (217195,162409,249332,87980) by looking at the source. Even then, I didn't find all the holes by reading through large amounts of source. I found 87980 by investigating an error I saw in the JS console during normal use.
I found 162409 during John Keiser's presentation about a feature he maintained. His slide said something like "Session history uses keys to recognize form controls when you return to a page:
Tagname>InputName>InputType>FormName>Index
I'd probably find more code-level holes if I spent more time looking at source code
Only five of the ten security holes involved memory management (buffer overflows, etc). Two were file permission problems and three were logic errors. So avoiding manual memory management would only have kept out half of the holes.
Most of Mozilla uses safer abstractions such as reference-counted pointers and string classes (which are, of course, implemented using manual memory management). But Mozilla manages memory manually in a few places where speed matters a lot, such as image rendering. I think that's the right strategy, even though it occasionally introduces security holes, because overall speed relative to other browsers is important.
The delay isn't an insulting, newbie-proofing feature to "make sure you read the dialog". It's a fix for a real security hole: a race condition involving human reaction time. Because it is a fix for a real security hole, there should not be a setting in about:config to reverse it. I'm disturbed that I haven't seen fixes for the same security hole in Opera and Internet Explorer.
If you know that your worst-case reaction time is less than two seconds, you can change the delay.
I listed it as a "major bug fix" in the changelog I created. That does not mean that it was treated as a "major bug" before it was fixed, only that I thought it was a notable fix (at least for the kinds of people who would read a changelog).
("Effect" as a verb means "to bring about or execute".)
Maybe the post meant "0.9 must be ported, or I will mark Firefox (0.8) as broken due to security issues". I'm not familiar with the process and jargon surrounding distributions, so I can't be sure.
Look up the "halting problem" sometime. It's impossible to algorithmically prove that a computer program is correct (bug-free, etc.).
It is entirely possible to write proofs that programs are correct. In the Advanced Algorithms class I took, many of the homework problems were proving that short pseudocode programs were correct. (Proofs can be verified algorithmically, but that's not important the debate over whether it possible for software to not have security holes.)
You misunderstand the halting problem. What is impossible taking an arbitrary program as input and always being able to say either "This program halts" or "This program does not halt". But if you're also allowed to say "I don't know" for some input programs, it is no longer impossible.
Here's a dumb example:
if (program == "")
return HALTS;
else if (program == "while(1);")
return DOES_NOT_HALT;
else
return DONT_KNOW;
Btw, there's something called "proof-carrying code", but I don't know much about it.
Firefox uses "I'm Feeling Lucky", which always goes straight to the first hit. IE with the Google Toolbar uses "Browse By Name", which only goes to the first hit when Google is sure the first hit is the correct one. Otherwise, it displays search results.
You can make Firefox use "Browse By Name" by changing a hidden pref. Or you can make it use a normal Google search if you want.
All joking aside though, I have no idea how people got anything done before the internet.
Why don't you search Google and find out?
MSIE support sucks. It is getting better, but it still sucks
I don't see how MSIE's lack of alpha-transparency could stop you from using PNGs, since you use GIFs now.
As far as I know, short of a dumb terminal, there's no system in the world that can prevent users from doing dumb things to their files. It's the ones that screw up the system that need to be prevented.
As you said, an operating system can't prevent users from doing dumb things to their files, but generally prevent users from doing dumb things to system files. On the other hand, a web browser should prevent web pages from doing things to users' files. When it doesn't, we call it a security hole in the web browser. On a system with only one user, this kind of hole is more severe than an operating system hole that lets the malicious code then gain root privileges and muck with system files.
I am jruderman@gmail.com and I have not had Gmail invites for a month. Any particular reason you're trying to get Slashdot users to spam me?
Your sig makes Slashdot invalid if you post twice in the same article.
I believe not serving the user would be an exception, rather than a rule.
Sure. But that's true of commercial software as well.
Sure, they might not prefetch pages, but if they did don't you think a lot of websites might start to simply block Mozilla, because it sucks up all their bandwidth?
Yes, I think that would happen.
What are you talking about it will not display the alt attribute without a title attribute? I assume that would be part of the rendering engine and this bug would be shared with Firefox? This doesn't occur in Firefox.
Go to www.alltheweb.com and hover over the logo. You'll get a tooltip in IE but not in Firefox, because the text is in an alt attribute. On alltheweb.com, it makes sense for there not to be a tooltip, but there are some sites that have crucial information in tooltips and incorrectly use the alt attribute instead of the title attribute to create the tooltip. The alt attribute is meant to replace the image when it is not shown (e.g. for blind users, lynx users, and users who have disabled images). Alternate text tends to make poor tooltip text and vice versa. Read bugs 25537,74241,41924 (or just Hixie's comments in those insanely long bugs) if you're interested.
Also, I don't know where you got your browser, but my copy of Firefox's help menu has "Release Notes" and "About Mozilla Firefox".
I'm using a branch nightly build and I see those items. I expect them to be there in Firefox 1.0 Preview Release.
While I can train a canonical "Mom" not to answer emails that say "Your bank needs to verifyt your mother's maiden name and social", I'm not so sure I can protect her against these other than to say "NEVER, EVER LOOK AT ANY EMAIL FROM A BANK OR EBAY". If you're running an auction, just keep going back to eBay for the info.
What's so hard about "Always check that you're actually on ebay.com after following a link from an e-mail"?
How does an ISP's DNS server become poisoned? Is it a weakness in a particular DNS server or in the DNS protocol?
It seems to me that, to be 100% sure, you must audit the source code line by line.
Given how long it takes most security holes in popular open source software to be found, I think we can safely say that nobody competent enough to audit the source of software they use does so.
And when it comes to that, Free Software is highly desirable, because it always tries to serve the user, above all other considerations.
Some examples where that is not true:
Mozilla does not display the alt attribute of an image as a tooltip when there is no title attribute. This serves blind users of all web browsers by encouraging web authors to use the alt and title attributes correctly, but it annoys some Mozilla users.
Mozilla does not pre-fetch every link on pages visit. While this would benefit users (at least those who don't pay for bandwidth), it would have a negative impact on web sites.
Mozilla supports autocomplete="off" (disables password manager for a specific form) and Cache-control: no-store (disables cache for a specific page), primarily due to blackmail from banks that would block Mozilla otherwise.
Firefox's Help menu contains "Tell a Friend" and "Promote Firefox" menu items. These support the Mozilla Foundation's marketing goals; they are not there to benefit users.
Firefox will not import your home page from Internet Explorer.
AIM is advertising-supported software because it displays its ads in the AIM window.
Kazaa is adware because it displays its ads while you use Internet Explorer. Pop-up adware often makes it difficult for users to tell what application the ads sponsor, which IMO is the point at which it becomes evil. I don't know whether Kazaa's ads say "This ad is shown using Claria technology to sponsor your use of Kazaa. To stop seeing these ads, uninstall Kazaa". I'm not going to install Kazaa to find out.
How many visitors does that site get a day?
In Deep Fried Live, you can click on many of the cooking tools and foods Tako uses during or after the cartoon.
Hmm, I didn't realize "Open Source" had to be capitalized.
If one had the technology to vary the intensity of red, green, and blue over an infinite set of real values, then RGB would be able to perfectly replicate any color.
Wrong. Take a look at a CIE Chromaticity diagram and you'll see that no matter what three wavelengths you choose as your primary set, there will be some colors you can't mimic.