Every extension should run in a separate process

[gordebil]

I am using Firefox with a fair number of extensions, but it has become significantly slower, and Firefox's task manager says the extensions are to blame. One or more extensions are consuming a massive amount of cpu cycles and memory. However it’s hard to tell which one is responsible because all the extensions share a single process. It's a hassle to disable and test each extension one by one. If Firefox had this feature, it could warn users about extensions that are too resource intensive and even limit them before the browser slows down completely. It might also make it easier to implement extension sandboxing in the future.

[Jon]

Thanks for submitting an idea to the Mozilla Connect community! Your idea is now open to votes (aka kudos) and comments.

[NARBM]

I think this move is also good for future-proofing Firefox, as more and more people will use extensions to get their view of a "usable web experience", and there may be a buggy extension used causing Firefox to hang or to have some other issue.  Pretty cool idea!

[DiverDown]

If you truly want no bleed between apps every processes should be isolated. It's a very doable thing if you take the right hardware approach. I understand most are used to working for a software side but the opposite is much better. I hope I can help others learn more about this.

[DiverDown]

I found out the hard way extensions are great but if you use to many they will choke your system out not necessarily because it can't be done but it's next to impossible in the current eco system, unfortunately changes need to occur before real progress is possible. 

[NARBM]

I agree that a good hardware platform with the right software can make a more secure experience.

In fact, the whole O.S. can be sandboxed (and, as a bonus, have the system files be replaced with fresh ones every reboot), and the firmware could only talk to the bootloader of a personal computer... later being set aside when the O.S. loads.  The problem is that security isn't as well-practiced as one should hope, as there are many misconceptions to security, like encryption being a universal hallmark of excellent security.  Security is a vast, polarising practice, but encryption isn't the only key to security.  If the keys for encryption standards are easily accessible, and security would only be powered by encryption, well... data goes goodbye, for example.

I think a good way for better hardware security to happen is with a standard RISC-V PC type (similar to the IBM PC).  Unlike the ARM architecture, which doesn't really have device standards and is a wild west with various devices using various standards, and for x86 with both Intel and AMD having system managers (Intel ME and AMD PSP) with various uses (that most have no need or want for), with potential vulnerabilities related to hijacking a PC remotely on the firmware level (not so much with AMD PSP, as it is considerably less privileged, but certainly with the Intel ME), RISC-V is in a very clean slate, and is a very experimental platform currently.  Only embedded computers and some niche PCs use RISC-V CPUs.

Maybe some engineers would make a much more stable RISC-V platform that will just be the standard for all desktops and laptops for the average Joe.  Or, maybe not.