Will you ever be able to do an upgrade that doesn't f&ck up all the extensions?

this upgrade doesn't "f&ck" up anything. you just have to change the maxversion.

So there's even less reason to let extensions die unless their authors update all of them? Backwards compatibility, anyone? If the FF 1.5 extensions API hasn't changed, can't it just run an extension despite a lower maxversion - you say it would work anyway.

This new release actually breaks some extensions.
Usually the more complex ones. (But no rule without exception :p)

What does not mean it will break all extensions.

"Active development" extensions will sooner or later receive updates, so that won't be a problem (at least after some time).

Like with 1.0 there will be for sure quite a bunch of "maxversion-modded-by-somebody-else" extensions available from different mirrors.

So if an extension is not updated anymore by the original author you may get one of those or mod it yourself.

Some extensions will die for sure.
But IMO thats better than that crappy "backwards-compatiblity mess" e.g. MS left with the WinAPI. :p

What do you mean "rev the extension version"? The point of the internals is to be able to update the max version on UMO and users don't have to download a new version. In which case we MUST get bug 305162 and bug 304857 fixed!

"you just have to change the maxversion"

Yeah, yeah...

As a 'regular user' I expect that if it not going to break my extensions, they should just work on an upgrade. If they are going to break then they should be disabled. I know, I know, how do you tell if an extension will break? That's up to the devs to figure out.

From an end user standpoint, the fact that I have 10 extensions installed that all need to have a text file edited and a version updated just to work, is annoying as hell. I know how to manually go update those files but my girlfriend doesn't want to have to know that.

Saying that extensions not under active development will die is stupid. If an extension is finished, it's not under active development. Active development generally means fixing bugs or adding features but many small extensions don't have any more bugs or new features to add. To expect the author to keep bumping versions for eternitiy is not practical.

Is it possible to mark an extension as compatible with 1.4 and 1.6a without also marking it as compatible with 1.5?

"Active developement" just means that the author(s) are around to maintain their products...
And that means that they'll test the extension with that new FX version and then release a new version that required maxversion value.

You (or that 'Regular User') expect extensions not to break.
Right.
Devs actually try to make this possible for x.x.x releases.
Actually it failed (on some extensions) in the past, but usually people learn from errors ;)

The upcoming FX 1.5 is a major release.
People may still except that everything works fine. But that is just wishful thinking.

Look at other software products and you'll notice that almost every major product version breaks at least some parts of backwards-compatiblity.

Especially when Devs improve massively improve part of the application...

On FX 1.5s part most breakage will be there due to security enhancements. (that native wrapper stuff...)

Take WinXP SP2 as an example.
Pretty popular product... Pretty much manpower spent on developing it... and done under MSes "you're apps will still work" promise they usually made in the past.
And what happend: Many real popular applications stopped working after patching SP2 into the system.
e.g. ICQ (IIRC)

The thing the FX drivers should do is to motivate devs to test and update their extensions before FX1.5 stable is released.
So that Update Checker will be able to fetch new versions right after running new FX the first time.
And guess what?: Asa is just doing exactly this with this post ;)

Can someone help me out when I load Firefox 1.5 Beta it just sits on the Testing the Capabilities of your extensions screen, then I cancel it and load firefox and non of my extensions are working. Whats the fix???

Personally I'm tired of "repairing" extensions with every new version of Firefox.

I already bumped the maxVersion values on UMO for most of my extensions. Beta 1 did not disable them upon upgrade.

The beta is a chance to get the extension situation up to scratch. Beta 2 is on the way before Firefox 1.5, so it's not like the community is short of time to fix things that need fixing.

"Look at other software products and you'll notice that almost every major product version breaks at least some parts of backwards-compatiblity."

As far as I'm concerned, that's a completely unacceptable argument. Major software products take *huge* pains to avoid pissing off customers if something is broken. I've worked for Sun and have witnessed first-hand how seriously they take the issue. That's why Solaris has an excellent reputation for stability/backwards-compatability. Old software just works. Period. You mention XP, and it's true that a few apps had problems. But it's also true that they made huge efforts not to break lots of other stuff. And, if you'll note, even the small amount of stuff XP broke caused lots of users to get (vocally) upset. How many still won't touch SP2 (even with the better security) because they're afraid it will break something?

Firefox gets a bit of slack simply because it's hard for a new environment to "find its place" and not break things in the process. But it also claims to make life simple for the average user... Causing every extension to break (until the author hopefully gets around to updating a version string) is decidedly a PITA for average users. There's wiggle-room because the pending release is a Beta, but you can be sure there will be lots of extensions not updated by the time 1.5 Final is released.

Blame the Mozilla/extension developers all you want, but it's the users that suffer in the end.

Firefox is still relatively new in terms of maturity. Desirable API stability just isn't there, considering the way that extensions work and it won't be for some time if Firefox is going to continually improve.

In fact, there is no API at all. Extensions interface directly with Firefox code. It's relatively difficult (if not impossible) to maintain a high level of backwards compatibility, yet still offer new features or bugfixes. As Firefox gets older, the number of changes (and thus breakages) should be fewer.

The bottom line is this: You all have the power to improve this situation (by testing, contacting authors, making sure that abandoned projects get new owners), so if you have 10 minutes to spare per day, do it! MozillaZine has a perfectly good extensions forum that's been horribly empty of late. Go there, start a thread, get people interested, make things happen.

The bottom line is also this: Most of my extensions still work without further modification (other than a remote maxVersion bump). This will be the case with many, many others as well.

I, and I think most others, agree wholeheartedly with Justin. I just don't understand why it is such a hard concept for Firefox developers to grasp. The common every day user is going to get sick of this real quick (if they haven't already) and will head back to IE. Issues such as this are major obstacles in the mainstream acceptance of Firefox. I really want Firefox to succeed but illogical decisions like this aren't helping our cause.

You ppl whine to much... Firefox is still a very young application... If you want to be backwards compatible, you can't make a giant leap in features and or security (or anything else for that matter).

"Firefox is still relatively new in terms of maturity. Desirable API stability just isn't there"

Well, yes and no.

It inherits large chunks of Mozilla, so it's not all brand new. One certainly shouldn't expect Solaris-like API stability yes, but the "break all extensions every 6 months" ought to raise an eyebrow.

There are APIs that are supposed to be stable ("frozen"), although AFAIK that only applies to the C++ APIs. At least, as an extension developer I've never seen any reference to JS API stability levels, or any way to check to see if I'm compliant. I think that's what's really needed, at some point, to help this situation. The versioning needs to be finer-grained than a single global value. That would help prevent needlessly breaking extensions, and it would help developers know what to expect. It's also says that Mozilla.org has the confidence to stand behind that API for duration appropriate for the stability level.

"If you want to be backwards compatible, you can't make a giant leap in features and or security (or anything else for that matter)."

Sure you can, you're creating a false dichotomy. One just needs to manage potentially incompatible changes in a suitable way. Yes, it's sometimes harder than just throwing compatibility out the door. But developers anywhere want to see appropriate commitment to API stability. Especially when one person's code change can affect hundreds of developers.

Ideally, the end user won't have any problems to get sick of. They'll be waiting for 1.5 stable, or maybe beta 2, which should give most extension developers time to update their extensions and/or compatibility info. All the end user will get is the "upgrade your extensions" notice on upgrading FF.

Yes, it would be better if that wasn't necessary either. Yes, it's a pain that the extension developers have to bump the compatibility info. Yes, there is a problem lying in wait over extensions that won't get maintained. But the average user is not going to be stuck with a dozen extensions that break until he edits a bunch of text files.

Backward compatibility is a good idea, and something that I think FF should look into for 2.0, but let's be realistic about what most people are going to be doing. Right now, the extension/upgrade problem is mainly an issue for extension developers and beta users -- not for the common everyday user.

Also... if you want, you have the option of using only the frozen interfaces from JS, if you really want to be backwards-compatible. Many were frozen in Mozilla 1.0. They should all still be quite usable, and it is always an option to restrict oneself to only these interfaces.

If it sounds silly to limit oneself in this manner (many of the interfaces are outdated or do not give the specific functionality that you may be looking for), it's because Mozilla has advanced so much since then. However, using the more advanced interfaces comes at the cost of possible backwards incompatibility.

To check if you are "compliant", make sure that the interfaces you query from JS are frozen in the C++ IDLs. I think a lot of the DOM is, and probably some of the plugin architecture and other APIs designed for third-party embedders.

In short... you could set your version for 1.9 right now, if you knew you were only using the frozen APIs, and you would have pretty much guaranteed backwards compatibility on your side. If you don't restrict yourself to this, you have to be prepared to accept that there might be some updates required to your "finished" extension.

To Dave, who experienced endless extension check testing,

I found if I enabled Norton Internet Security (2005 Anti-Spyware Edition, Windows version), Firefox could not finish extension checking. After disabling the Norton temporarily, Firefox started normally.

People having extension problems can use Nightly Tester Tools and the "Make Compatible" option when right clicking an extension. Works awesome!

Is there a shortage of reviewers on UMO? I've had a max-version bumped extension sitting in the approval queue for about three days now :p

"It inherits large chunks of Mozilla, so it's not all brand new. One certainly shouldn't expect Solaris-like API stability yes, but the "break all extensions every 6 months" ought to raise an eyebrow.

There are APIs that are supposed to be stable ("frozen"), although AFAIK that only applies to the C++ APIs."

Extensions *use* C++ APIs, sure, but they work by doing things with the interface. The interface isn't stable, and as I said, there's a choice between making necessary fixes/improvements and retaining backwards compatibility with extensions that rely too much on the inner-workings.

Take for example, extensions that hooked into the previous Options dialog. There's no way that consistency could remain between the old and new dialog at all. The extensions (TBP) using it had to be updated. Likewise, the drag/drop tab addition meant that some extensions had to change.

Most extensions require few (if any) changes, but sometimes the only way to achieve certain tasks is with dirty hacks, and dirty hacks always break eventually.

Anyway, some of my extensions have been barely modified (if at all) since 0.8. It's not like everything breaks with every new release.

The only way I see to address this is to give most elements within the interface a version number and apply version checking on a much wider scale. Stuff like the context menu hasn't fundamentally been altered since Firefox 0.7, but tabs have.

It doesn't just break extensions. Worse yet, extensions break it. If you have installed extensions and you update Fx, there is no assurance that the new version will work.

With version 1.5 they are attempting to improve the update process. Unfortunately, as far as I can tell, they are doing this by a process of trial and error. Let's try all these release candidates and see what breaks. Let's see if it still seems to work after updating.

Unfortunately, users are encouraged to extend Firefox, but there is no assurance of backward compatibility, and there is no warning to that effect. Even worse, there is no assurance of the safety of the extensions. In my opinion it should not be presented to the public at all for use with extensions.

"Firefox is still relatively new in terms of maturity."
"You ppl whine to much... Firefox is still a very young application..."

Hellloooo! Firefox has been developed for over 3 years! And it's based on the Mozilla Gecko code, which is *much* older. It's hardly a new application, the "1" version number doesn't mean anything. So stop the stupid excuses. This wouldn't be a problem if Firefox wouldn't rely so much on 3rd party extensions...

And it's based on the Mozilla Gecko code, which is *much* older.

What does that have to do with the maturity of the extensions system, which didn't exist in Mozilla?

Beep - wrong! Extensions do exist in Mozilla Suite, though it's not the main feature of the browser.

But to answer your question: they had a lot of time (over 3 years!) to improve extension API, because they didn't write Firefox from scratch.

I would argue that Firefox relies on extensions, and that ordinary, average, common, normal, end users have problems with extension compatibility.

It's been my overall observation that these types of users have little to no extensions installed, the default theme and have barely looked through the options. If they are using a customized or extended version, it's been installed or is maintained by a more experienced individual.

In most of the installs I've personally done or encouraged, when I check back in a few months the only thing different would be the personal info and cache!

Those that want more do for themselves, either by scouring UMO and TEM for extensions or by authoring their own solutions. The "penalty" paid is that stability/compatibility is compromised when running anything less than release versions of Firefox or an extension.

I was falling in love with the nightly builds just the other day, until I hit the roadblock of incompatible extensions, specifically in my case the roboform toolbar... I simply "can't" live without it, so back to 1.06 for me. This represents the real issue with compatibility, losing the functionality an extension provides while "testing" a recent build.

This is simply usage far beyond the mythical average user, so is there a need to protect them while disproportionately affecting development?

rtk, I would argue that Firefox doesn't rely on extensions. The most popular extension has only been downloaded in the hundreds of thousands of times. Firefox has been downloaded nearly 90,000,000 times. You can do the math on how much we rely on extensions :-)

- A

Asa why are you still quoting the number of downloads instead of the installed base? You aren't helping your cause by cooking your numbers.

Asa, despite the wording of my message, I was agreeing with you. Most installs I come across (that I haven't done myself) have no extensions installed. My point was simply that those that use extensions expect there to be issues when upgrading Firefox, and have the resources available to deal with the inevitable compatibility issues. As well, development of Firefox shouldn't be based on protecting a user that simply doesn't exist (an "average" user with multiple extensions installed).

I rely on extensions, but I don't believe "we" do. ;-)

"The most popular extension has only been downloaded in the hundreds of thousands of times. Firefox has been downloaded nearly 90,000,000 times."

I propose removing extension support in the 1.5 final release -- along with the Javascript Console -- as both are clutter that confuse and frighten average users. Users who want either feature can just install it as an extension.

(tongue in cheek :-)

"The most popular extension has only been downloaded in the hundreds of thousands of times."

Asa,

I don't know anyone that uses update.mozilla.org much if that's where you're getting those numbers from. extensionsmirror.nl is where I always go as they have a larger selection of extensions.

Plus, you have to add up the numbers from mozdev, extenionsmirror, update.mozilla.org, all websites that have links to extensions that are not on one of those sites. I'm not saying that adds up to (the inaccurate) 90-million number, but it's going to be much higher than update.mozilla.org.

Plus, you can't look at the download numbers for an individual plugin. You have to consider the total number of people that have any single extension installed. Basically, the union of {all people that have extenions A installed} , {all people that have extension B installed}, etc. etc. That number is going to be in the millions for sure. (I personally don't know anyone that have no extensions installed, and any friends of mine that set new users up with FF always install some subset of thier extensions for the new user)

I don't think FF should maintain perfect backwards compatability, but the current compatability check of a version number in the extensions dir is really bad. Maybe FF should provide a script for each extension to use to check for compatability on startup or something, I don't know. I'd have to sit down with a few people and brain storm on solutions.

(btw, I'm still using FF 1.0.4 because after all the headaches of extensions breaking from 1.0.3->1.0.4 [which were't true incompatabilities] I'm not willing to go make all the version changes to support a 1.0.6 upgrade -- and neither are many users I kow.)

As far as extension use goes, I can’t help but think that the primary user of them would be the more “sophisticated” user (i.e. experienced).

The average user doesn’t fit into this category. In fact the average user almost never downloads any software, period.

If that average user is using Firefox, it’s because their techie friend talked them into it and then installed it for them as well.

If Opera were free it would be kicking Firefox’s butt, as far as user base goes.

One advantage to Opera is it has far more options and functionality than Firefox does (sans extensions). Now you can equal or exceed Opera functionality in Firefox, provided you hunt down and install all the right extensions.

I realize one of Mozilla’s goals with Firefox was to make it “lean and mean out of the box” however, with the exception of tabbed browsing, there’s not much else to it (to the average user) unless they are willing to install extensions.

With that in mind, I found that Opera ran much faster than Firefox on Windows XP (when Firefox had extensions installed to equal Opera’s functionality, and then some).

I still prefer the UI of Firefox and I am a fan of Open Source, but Mozilla has a way to go to get Firefox to where I think it should be.

It seems to me that they could include the functionality of the most popular extensions “out of the box” and still have the best browser out there (in fact, better than what they have now).

I can’t help but think it would run better with that stuff written into the code rather than patched on afterwards by Joe Coder, Ace l33t and others who don’t have the benefit of a QC department.

A number of extensions have really screwed up my Firefox installations over the past 18 months.

On another note, the current setup with extensions does Mozilla a big disservice with every new release of Firefox. Every time a new version of Firefox is released it’s incompatible with most previously installed extensions AND the extension authors are slow to update. I’ve typically taken to ignoring any “hoopla” around a Firefox release and just pretended it was a few weeks later than it actually was.

That way, I get updated (or any, period) extensions at the same time I get the new version of Firefox.

And why does http://extensionroom.mozdev.org/ exist?

As far as extension use goes, I can’t help but think that the primary user of them would be the more “sophisticated” user (i.e. experienced).

The average user doesn’t fit into this category. In fact the average user almost never downloads any software, period.

If that average user is using Firefox, it’s because their techie friend talked them into it and then installed it for them as well.

If Opera were free it would be kicking Firefox’s butt, as far as user base goes.

One advantage to Opera is it has far more options and functionality than Firefox does (sans extensions). Now you can equal or exceed Opera functionality in Firefox, provided you hunt down and install all the right extensions.

I realize one of Mozilla’s goals with Firefox was to make it “lean and mean out of the box” however, with the exception of tabbed browsing, there’s not much else to it (to the average user) unless they are willing to install extensions.

With that in mind, I found that Opera ran much faster than Firefox on Windows XP (when Firefox had extensions installed to equal Opera’s functionality, and then some).

I still prefer the UI of Firefox and I am a fan of Open Source, but Mozilla has a way to go to get Firefox to where I think it should be.

It seems to me that they could include the functionality of the most popular extensions “out of the box” and still have the best browser out there (in fact, better than what they have now).

I can’t help but think it would run better with that stuff written into the code rather than patched on afterwards by Joe Coder, Ace l33t and others who don’t have the benefit of a QC department.

A number of extensions have really screwed up my Firefox installations over the past 18 months.

On another note, the current setup with extensions does Mozilla a big disservice with every new release of Firefox. Every time a new version of Firefox is released it’s incompatible with most previously installed extensions AND the extension authors are slow to update. I’ve typically taken to ignoring any “hoopla” around a Firefox release and just pretended it was a few weeks later than it actually was.

That way, I get updated (or any, period) extensions at the same time I get the new version of Firefox.

And why does http://extensionroom.mozdev.org/ exist?

As far as extension use goes, I can’t help but think that the primary user of them would be the more “sophisticated” user (i.e. experienced).

The average user doesn’t fit into this category. In fact the average user almost never downloads any software, period.

If that average user is using Firefox, it’s because their techie friend talked them into it and then installed it for them as well.

If Opera were free it would be kicking Firefox’s butt, as far as user base goes.

One advantage to Opera is it has far more options and functionality than Firefox does (sans extensions). Now you can equal or exceed Opera functionality in Firefox, provided you hunt down and install all the right extensions.

I realize one of Mozilla’s goal’s with Firefox was to make it “lean and mean out of the box” however, with the exception of tabbed browsing there’s not much else to it (to the average user) unless they are willing to install extensions.

With that in mind, I found that Opera ran much faster than Firefox on Windows XP (when Firefox had extensions installed to equal Opera’s functionality, and then some).

I still prefer the UI of Firefox and I am a fan of Open Source, but Mozilla has a way to go to get Firefox to where I think it should be.

It seems to me that they could include the functionality of the most popular extensions “out of the box” and still have the best browser out there (in fact, better than what they have now).

I can’t help but think it would run better with that stuff written into the code rather than patched on afterwards by Joe Coder, Ace l33t and others who don’t have the benefit of a QC department.

A number of extensions have really screwed up my Firefox installations over the past 18 months.

On another note, the current setup with extensions does Mozilla a big disservice with every new release of Firefox. Every time a new version of Firefox is released it’s incompatible with most previously installed extensions AND the extension authors are slow to update. I’ve typically taken to ignoring any “hoopla” around a Firefox release and just pretended it was a few weeks later than it actually was.

That way, I get updated (or any, period) extensions at the same time I get the new version of Firefox.

And why does http://extensionroom.mozdev.org/ exist?

As far as extension use goes, I can’t help but think that the primary user of them would be the more “sophisticated” user (i.e. experienced).

The average user doesn’t fit into this category. In fact the average user almost never downloads any software, period.

If that average user is using Firefox, it’s because their techie friend talked them into it and then installed it for them as well.

If Opera were free it would be kicking Firefox’s butt, as far as user base goes.

One advantage to Opera is it has far more options and functionality than Firefox does (sans extensions). Now you can equal or exceed Opera functionality in Firefox, provided you hunt down and install all the right extensions.

I realize one of Mozilla’s goal’s with Firefox was to make it “lean and mean out of the box” however, with the exception of tabbed browsing there’s not much else to it (to the average user) unless they are willing to install extensions.

With that in mind, I found that Opera ran much faster than Firefox on Windows XP (when Firefox had extensions installed to equal Opera’s functionality, and then some).

I still prefer the UI of Firefox and I am a fan of Open Source, but Mozilla has a way to go to get Firefox to where I think it should be.

It seems to me that they could include the functionality of the most popular extensions “out of the box” and still have the best browser out there (in fact, better than what they have now).

I can’t help but think it would run better with that stuff written into the code rather than patched on afterwards by Joe Coder, Ace l33t and others who don’t have the benefit of a QC department.

A number of extensions have really screwed up my Firefox installations over the past 18 months.

On another note, the current setup with extensions does Mozilla a big disservice with every new release of Firefox. Every time a new version of Firefox is released it’s incompatible with most previously installed extensions AND the extension authors are slow to update. I’ve typically taken to ignoring any “hoopla” around a Firefox release and just pretended it was a few weeks later than it actually was.

That way, I get updated (or any, period) extensions at the same time I get the new version of Firefox.

And why does http://extensionroom.mozdev.org/ exist?

As far as extension use goes, I can’t help but think that the primary user of them would be the more “sophisticated” user (i.e. experienced).

The average user doesn’t fit into this category. In fact the average user almost never downloads any software, period.

If that average user is using Firefox, it’s because their techie friend talked them into it and then installed it for them as well.

If Opera were free it would be kicking Firefox’s butt, as far as user base goes.

One advantage to Opera is it has far more options and functionality than Firefox does (sans extensions). Now you can equal or exceed Opera functionality in Firefox, provided you hunt down and install all the right extensions.

I realize one of Mozilla’s goal’s with Firefox was to make it “lean and mean out of the box” however, with the exception of tabbed browsing there’s not much else to it (to the average user) unless they are willing to install extensions.

With that in mind, I found that Opera ran much faster than Firefox on Windows XP (when Firefox had extensions installed to equal Opera’s functionality, and then some).

I still prefer the UI of Firefox and I am a fan of Open Source, but Mozilla has a way to go to get Firefox to where I think it should be.

It seems to me that they could include the functionality of the most popular extensions “out of the box” and still have the best browser out there (in fact, better than what they have now).

I can’t help but think it would run better with that stuff written into the code rather than patched on afterwards by Joe Coder, Ace l33t and others who don’t have the benefit of a QC department.

A number of extensions have really screwed up my Firefox installations over the past 18 months.

On another note, the current setup with extensions does Mozilla a big disservice with every new release of Firefox. Every time a new version of Firefox is released it’s incompatible with most previously installed extensions AND the extension authors are slow to update. I’ve typically taken to ignoring any “hoopla” around a Firefox release and just pretended it was a few weeks later than it actually was.

That way, I get updated (or any, period) extensions at the same time I get the new version of Firefox.

And why does http://extensionroom.mozdev.org/ exist?