Slashdot Mirror
Firefox 44 Deletes Fine-Grained Cookie Management (mozilla.org)
ewhac writes: Among its other desirable features, Firefox included a feature allowing very fine-grained cookie management. When enabled, every time a Web site asked to set a cookie, Firefox would raise a dialog containing information about the cookie requested, which you could then approve or deny. An "exception" list also allowed you to mark selected domains as "Always allow" or "Always deny", so that the dialog would not appear for frequently-visited sites. It was an excellent way to maintain close, custom control over which sites could set cookies, and which specific cookies they could set. It also helped easily identify poorly-coded sites that unnecessarily requested cookies for every single asset, or which would hit the browser with a "cookie storm" — hundreds of concurrent cookie requests.
Mozilla quietly deleted this feature from Firefox 44, with no functional equivalent put in its place. Further, users who had enabled the "Ask before accept" feature have had that preference silently changed to, "Accept normally." The proffered excuse for the removal was that the feature was unmaintained, and that its users were, "probably crashing multiple times a day as a result" (although no evidence was presented to support this assertion). Mozilla's apparent position is that users wishing fine-grained cookie control should be using a third-party add-on instead, and that an "Ask before accept" option was, "not really nice to use on today's Web."
Mozilla quietly deleted this feature from Firefox 44, with no functional equivalent put in its place. Further, users who had enabled the "Ask before accept" feature have had that preference silently changed to, "Accept normally." The proffered excuse for the removal was that the feature was unmaintained, and that its users were, "probably crashing multiple times a day as a result" (although no evidence was presented to support this assertion). Mozilla's apparent position is that users wishing fine-grained cookie control should be using a third-party add-on instead, and that an "Ask before accept" option was, "not really nice to use on today's Web."
Seems to be as fine grained as I need.
Ah, I see they are following the Gnome school of user interface design.
They seem to be really trying to shoot themselves in the foot lately.
I have an add-on that keeps only the cookies I explicitly select, the rest get deleted whenever I close Firefox, or when I manually delete cache and cookies with shift-control-delete. Just get that and have all the 'fine-grained' control you want.
Are YOU using the TOOL, or is the TOOL using YOU? Think about it!
I leave a site, its cookies explode.
Moreover, the allegation that enabling the feature destabilized the browser is pharmaceutically pure bullshit. I've been using the feature since its inception, and have Firefox windows open and running for days at a time without ill effect.
Contrariwise, I just went to check my cookie store, and found a bunch of new, unapproved, unwelcome, provably unnecessary cookies have appeared in just the week since I moved from v43 to v44. Deleting them after the fact is not a solution. Once set, tracking can take place immediately. The damage has already been done.
The proffered reasons for the change are easily shown to be false, so I do not hold out any hope that Mozilla management will have a change of heart on this matter and reinstate the long-standing feature.
Would anyone care to recommend a cookie management add-on?
Editor, A1-AAA AmeriCaptions
I built a new Windows image for our workstation PXE deployments, this time without Firefox.
If you're going to be just another trash browser you're no longer getting installed on the systems I'm responsible for.
In true Mozilla fashion, the discussion on the bug tracker has been censored, so people can't even effectively complain about it.
You are mostly right. Although it is unclear how many other extensions won't be adaptable to the new model. They are working with NoScript because NoScript is the 3rd most popular add-on for firefox. But what about those odd-ball add-ons that only have a couple of hundred users?
Meanwhile one thing that is legitimately and inarguably stupid is this add-on signing requirement they keep pushing back every couple of releases. They want to force you to submit your add-on source code to them for signing. At first they were doing automated code inspection and rejected add-ons that didn't pass, even for stylistic reasons. It took a couple of months of bitching before they finally backed off that level of scrutiny, doh!
But it is still a problem for people who have internally developed extensions - forcing them to choose between running an unsupported version of firefox or exposing their source code to mozilla who can not guarantee that it won't be pilfered away via corporate espionage.
All they need to do to fix it is make mozilla check for a list of exceptions to the signature requirements in an admin-only writable location (like /usr/lib/mozilla/ on linux or an admin-only part of the registry on windows). The code to do that is already 99% written because they already pull config data out of those locations, just need to verify it is admin-only writable.
But they keep resisting the obvious, instead insisting that anyone who wants to run an unsigned add-on must run a completely separate installation of firefox and thus forgo all the security benefits of getting auto-updates straight from them. The end result is much reduced security for those people - no crypto signatures for any add-ons and they must do manual compiles each time there is a new firefox release - and really, only the most hard-core of users is ever going to do that in a timely fashion. Just because you have an odd-ball add-on doesn't mean you are that hard-core.
I'm not that hard-core, but I still run the defunct "redirect cleaner" because none of the replacement add-ons quite match the original's functionality in corner cases. If I had enough time to compile every new release of firefox, I would have enough time to fix one of the replacement redirect-cleaner extensions to handle the corner-cases too.
Yes, I "shouted". Obviously to OP has no clue.
Denying the creation of a cookie in the first place has nothing to do with deleting them when Firefox is closed (whoever closes ALL of their FF windows anyway?).
I hope Pale Moon keeps the feature, but, IMO, FF44 is now nearly useless.
Some where back in the dim recent past, Firefox's ESCape key no longer meant abort everything and return control completely to the user. No matter if the base html is incomplete, no matter if some goofy-gumdrop JSON cloud-abortion is in progress, or a 302 redirect is in progress. No matter if you'll have to settle for a blank page because CSS cannot decide what color the text will be. Just ABORT. Now the ESC key means hardly anything.
Now in the face of incomplete loads, packet loss, severely delayed DNS lookups, javascript tumors that are busy metastasizing to grow the page from seeds using repeated lookups to unresponsive and overworked database servers --- all of this results in pages that won't stop loading, tabs that will not close immediately, or even pages with visible readable content that will not respond to scrolling requests or link clicks... until... exactly what I never found out.
The purported reason was to save the poor deep data content providers from aborted transactions caused by unwashed masses hitting reload and ESC. I say, if they're overloaded or vulnerable in any way to aborts or identical re-submits they are vulnerable to script kiddies too and someone has not done their job properly or provisioned their servers adequately. I never considered the ability to abort a web load as anything but an intrinsic RIGHT --- until it was taken away. It was,like, what are they thinking?
I've had to force-close Firefox to regain control. And no we're not talking about Flash or embed delays either, I run NoScript. This is Firefox's native process refusing to abort everything under all conditions.
If content providers bite into some apple of complexity (for example) embedding advertising and load sharing schemes that do little tricks (such as) using gobblegook DNS names with low or zero TTL, they deserve to be sandbagged for their effort by the masses until they re-think their decision and (god forbid) roll back in the general direction of 'static' content.
Unfortunately this is something a third-party addon cannot really fix. If ever I was temped to fork a whole project and create a new subculture to fix one aggravating feature=bug this is it.
<blink>down the rabbit hole</blink>