Firefox 3.6 vs 8.0 | Hacker News Reader

Firefox 3.6 vs 8.0

zdnet.com

83 pointsmelling64 comments

Firefox 8 is light years ahead of 3.6 in performance and memory usage. It's light-weight and blazing fast. If you've been away for a while, delete your old Firefox profile and re-import your setting. You'll be surprised at how much better Firefox is.

Is that manual cleanup really necessary?

No, but many of the "Firefox crashes a lot" are caused by profiles who have gotten corrupted over the years, and many of the "it eats 2G of RAM" are caused by that and extensions.

Old Firefoxen also tended to fragment their SQLite databases heavily on disk (and in some cases, internally as well), which can show as sluggish response and "hanging" whenever disk I/O is involved. There were changes to ensure that some critical database files don't get fragmented any more, but if you already have those files from an old install you're SOL as it can't defragment them while its using them.

If you have serious performance or memory issues with Firefox, and you're a longtime user, at this point cleaning your profile is probably at least as likely to help as waiting for the next version.

Edit: Just to clarify, the manual cleanup is necessary because uninstalling (and reinstalling) Firefox never deletes your profile. This means that a serious problem can never be fixed by a reinstall - a behavior that probably should be changed.

To defrag your databases, first close Firefox then run:

  for i in ~/.mozilla/firefox/*default/*ite; do sqlite3 $i vacuum; done

Note that this does not defragment the storage on disk, just the internal SQLite pages. So it's not a complete solution.

Actually, that command will also rewrite the entire .sqlite file, and if you have enough free space, the filesystem will most likely rewrite it in a contiguous manner.

I tested this with filefrag.

I've actually seen a Firefox developer suggest it on another HN thread, especially vacuuming your FF's sqlite3 database.

There needs to be a better solution than deleting your profile and starting from scratch.

https://addons.mozilla.org/firefox/addon/places-maintenance/ should do just fine?

In my opinion, Mozilla should have put more effort into seamless background updates before starting their 6-week release cycles. (Unleash the fanboys...)

I'm the Director of the Firefox Desktop product (I wasn't at the time of the move to the new release cadence) and I agree with you 100%. We've been busting our butts for the last couple of months to get our updates ironed out and smooth. We've already shipped some of this in Firefox, including not nagging users to restart, killing the "What's New" tab that would interrupt your start-up, and reducing the number of incompatible add-ons by helping users get rid of add-ons they never requested (add-ons that were sneakily installed without user consent.)

We've got "install add-ons in the background" code finished and in testing. We've got "bypass the UAC prompts" code finished and in testing. We've got add-ons defaulting to compatible with a blacklist for broken ones (reversing from a default to incompatible with a whitelist) coded and it's in testing. We'll be shipping these in Firefox 10 and 11 so the remaining bits are going to arrive very early in the new year.

This is awesome, great work. I hope unnoticeable updates become the norm for all desktop software.

I have to ask, when do you plan on dropping the version number from Firefox releases?

>I hope unnoticeable updates become the norm for all desktop software.

As long as you can turn that off, sure. I don't like software bypassing my package managers. Firefox is so far the sole exception I'm okay with since I'm using the Nightly builds and don't update as often. But for everything else, that would be a nightmare. I like to be in control of my updates, thank you.

It's time to let go and trust the developers of the software you're using to know when updates are stable, or when you really need a critical bug or security fix (if you don't trust them, why are you running their software anyway?).

Package managers are a great upgrade from having to manually update individual applications, and are still great for servers or other cases where upgrading can break crucial pieces of infrastructure - but for consumer desktop applications, there's no reason to waste your time doing sysadmin work.

When everything is a web app you'll have to get over it anyway. :)

For some things? Absolutely. For everything I currently use? The three hours I spent last week downgrading my Xorg thanks to a botched nvidia driver update force me to disagree.

Granted, it was an nvidia driver update; I should have known better. But that's exactly my point.

I don't get the point.

Was the nvidia driver update forced upon you and done behind your back? Or - did you do that intentionally and broke (by accident) stuff?

If the latter (and that's what it sounds like to me), isn't that totally unrelated to silent software updates and just anecdotal bad press about the nvidia driver and its stability?

He is saying something he wants to control the updates on, obviously the important stuff. The video driver is very important, but the browser is also very important, since a lot of people would not be able to get work done without it.

Exactly. I chose to update my video driver, so I knew what broke.

Imagine that update had happened automatically. Perhaps the day before a big deadline -- a stellar choice for disaster to strike! Now I'd have no clue what just broke and now I'd have to spend hours fixing it.

Intrusive updates shouldn't be automatic.

I don't see how package managers and developer pushed updates are incompatible concepts. Installing chrome adds a repo to my apt sources, it updates with the rest of my system, weekly and silently. I don't waste my time doing sysadmin work, and I get an up to date browser.

>It's time to let go and trust the developers of the software you're using to know when updates are stable, or when you really need a critical bug or security fix (if you don't trust them, why are you running their software anyway?).

I agree partially. My point is that you absolutely need the ability to turn it off. Always. But I'm fine with it being the default because that's what is most useful for the majority of users.

>When everything is a web app you'll have to get over it anyway. :)

You'll have to pry locally running programs out of my cold dead fingers. Web apps are fine for incredibly simple things. For everything else, they are as much buzzword bullshit bingo as the cloud.

- When you're in a bandwidth-limited situation like browsing over a tethered phone

- When you're using an access point operated by someone you don't trust

- When you want the best performance out of your system but need to have a browser open

- When any spike in latency could have bad consequences

- (possibly) when there's a significant chance of the computer losing power instantly and without warning

Five reasons I don't auto-update anything.

No way. A new release of my CAD software always screws something up. Unfortunately, it seems like some software is just too old and complicated to test thoroughly and cost effectivly.

That makes sense for security fixes and other internal issues. I would love an update that makes things faster. What I object to are auto-updates that change the user interface. In my opinion ANY interface changes should be user directed. If the new interface is better in some way, take the time and explain why.

EDIT: spelling

Already the version number isn't really advertised -- especially compared to Fx 1 through 4. However, it is an incredibly useful thing to know if you ever need support.

Version number shouldn't be dropped, however it should be, has been and will continue to be deemphasized.

Can you say anything about the "push" to 3.6 users? It was suppose to happen last week. I just checked on http://gs.statcounter.com and usage dropped .4% in the last week, but I'm not sure how accurate it is:

http://gs.statcounter.com/#browser_version-ww-daily-20111129...

I imagine that there are many 3.6 users who simply aren't into tech news like those of us here, and they simply don't know they should upgrade.

Nice work — that's definitely great to hear. Looking forward to the upcoming updates.

That's great to hear. Update-fatigue drove me away from Firefox. Chrome raised the bar with their seamless update system, I'm glad to hear that FF is up to meeting the challenge!

I'm really hoping you can get the remaining 3.6.x users to force-upgrade. We still have a couple people using 3.6 on our web app (where some pages won't render properly--we've labeled this 'wontfix' in our system since Firefox 7.0+ renders these pages properly.)

Any word on whether there is any way to get them to upgrade? Last I saw, it upgraded them to a newer 3.x version, but then opened a page on your site with a tiny-font-sized line of text that said "This isn't the latest version"--not exactly obvious that 3.6.x is pretty much ancient at this point.

Firefox 3.6 user here. I'd love to upgrade, but the new versions don't run on PowerPC

I'm not a PowerPC Mac user, but I understood PowerPC Firefox is now entirely handled by the community in the form of TenFourFox.

There are still extension issues causing me to wait. Also, I should probably start a fresh profile, and the effort is not something I'm particularly looking forward to.

Plus, the "Chrome wannabe" interface changes just piss me off. ;-)

Seriously? Heck, 3.6 isn't exactly a slouch when it comes to web standards, particularly compared to IE, Konqueror and some of the "smaller" engines. What features are you using that 3.6 won't render right?

Also, FF4 was released only 8 months ago, and FF3.6 received a security update less than a month ago. "Ancient" my ass.

I'm a Firefox user and I agree with you. Now, it's not horrible, but it should have been handled better. They might have lost the battle to Chrome in the long run just because of that.

(as Chrome has several advantages such as TV ads, billboards ads, included with all major software and often without an easy way to opt-out, ads on all major websites including the most major one, various lock-in attempts with major sites supporting stuff only Chrome provides, etc, while Firefox's advantage was just being there first)

Note that it's the above that makes me stick with Firefox - and also that lately it's been actually better than Chrome in many areas.

We've added millions of Firefox users since the move to the faster cadence. Another thing worth noting, we've been shipping updates about every 4 weeks from the beginning of the Firefox product's life. They were noisy and asked you to restart Firefox and prompted with UAC dialogs, and showed a progress meter when you were started. There's nothing new about all of that. We're probably showing that stuff less often now that we're on 6 weeks cycles and not 4 week cycles. We've also turned off the restart prompt for most users and the "What's New" tab for all users.

So, what's different now is that add-ons must move faster to keep up because we're not just shipping security and stability updates frequently, but actual feature updates that can break add-ons. Our big problem is not the noisy updates, but the add-on incompatibility. We're hard at work on that but it's a tough problem.

As you said some of the issues have been fixed already - and hopefully we can put all this in the past soon - but.

Whereas Firefox still gains users, it doesn't gain them as fast as the competitors and it seems to me that it is due to the bad press (even if from uninformed users, or from Google 'evangelists'). The bad press mainly came from the update issues, even thus some are gone now. That's the point I was raising.

I don't believe there are many people that will disagree with you on that matter. I read something today that described the release cycle as "increasingly erratic" and that was on a tech site which should know better. The improvements are there for all to see though, it's just a matter of whether users are willing to give each and every update a go.

Inaccuracies: SunSpider was not developed by Mozilla (it was Apple), and kraken is not based on SunSpider.

What about HTML5? A LOT of HTML5.

Firefox/Gecko has support for tons of HTML5 features that Chrome/Webkit doesn't support (or didn't until recently).

I wish some things would be pushed faster from Aurora into the other branches.

The only thing I dislike about Firefox is that there are a number of CSS1 features that still aren't supported. I just know Opera supports them, so I am not sure about the others. One can't style some form parts. There has been a bug report for this for ages.

Oh and to all web developers out there: Start checking for features, instead of browsers, PLEASE! It's so annoying to always read "Your browser doesn't support XY" when that's not true, because I am using a recent release, beta or the aurora branch of Firefox.

I made the 3.6 to 7/8 switch only a few months ago.

Here are some extensions I found critical to my sanity to keeping things somewhat familiar:

   Status-4-Evar   <<<<<<------------ a must
   Add-ons Manager Dialog Returns
   Back/forward dropmarker
   Restore View Source for Firefox 6+

and

   Firefox 3 theme for Firefox 4+ (v2.0)

Then once you are in FF8 treat yourself to extensions you can only running in FF8+ like Font Information which shows you what fonts are being used on the page in the page info dialog.

Now my FF8 looks a lot like 3.6 but works several times faster.

This has been fixed in Firefox 10 (or 9?). Until then you could use this extension:

https://addons.mozilla.org/en-US/firefox/addon/add-on-compat...

If you are responding to "only can be used in FF8" the font information extension only works in FF8+ because of the api hooks needed are not available in previous version. It's not a min/max versioning issues.

Link is broken (server taking too long to respond). Any alternatives to this URL?

Just carried out some Kraken 1.1 runs on Firefox on OSX, here on my iMac 8,1 (2.6ghz c2d/4gb ram).

I've got all major versions of Firefox installed side by side using my shell script (https://github.com/omgmog/install-all-firefox), this allowed me to test Firefox 2.0.0.20, 3.0.19, 3.6.24, 4.0.1, 5.0.1, 6.0.1, 7.0.1, 8.0.1, 10.0a2.

Results: http://omgmog.net/kraken.html

For Firefox 2.0.0.20 and 3.0.19 the browser kept crashing, so I was unable to finish these, but all of the other versions ran well.

Is it a coincidence that the last version of Firefox that will ever run on PowerPC macs is being surprisingly resilient to those calls to upgrade?

I'd be curious to see how many of the 3.6 laggards really are on old macs and are simply unable to upgrade until they budget in some new hardware.

Analytics for me says, for the last month:

  Firefox 8.0: 2.68% Mac
  Firefox 7.0.1: 9.53% Mac
  Firefox 3.6.24: 13.47% Mac

So it's definitely appreciably lower for 8 than for 3, but the closeness between 7 and 3 makes me wonder whether it's actually an effect.

So, what a Firefox 3.6 user is missing out on is running a bunch of artificial benchmarks faster?

If you haven't tried updating past 3.6, you should. Benchmarks may not be 100% accurate, but you can see the difference in JS-heavy apps (like GMail). The UI has been cleaned up and revamped. There are a ton of new HTML/CSS/JS features. Better memory management, and Firefox as a whole is faster. Restartless add-ons. The list goes on and on :)

Oh I have upgraded, and there are things that I like about it. However the title "What you are missing" isn't really addressed by showing a bunch of benchmarks. Having a real-world section talking about subjective performance on actual sites used by people would have done that.

So the only thing I get from Firefox 8 is faster javascript? Why is that useful when I would lose so much? I block all javascript with NoScript and go a few steps further with RequestPolicy and Ghostery.

Whereas I want my browser to look like a standard Windows classic theme window.

The title made me imagine it was between Firefox and V8

http://arewefastyet.com/

http://arewefastyet.com/?a=b&machine=8 probably more representative

http://arewefastyet.com/old-awfy.php is the most impressive

true, but i meant that most people reading HN are likely to be running the 64 bit version

Did Mozilla just pad their version number? Last I remember, there was Firefox 4 Beta. Why is it at 8 now?

They are now on a six-week release cycle, just like Chrome.

Thanks for your reply. Not sure why HN decided to downvote my question.

I just tried it. Guess what? The interface is still broken.

You look at the page contents constantly, at tab names often, at the bookmarks toolbar sometimes, at the URL occasionally, and you almost never look at the page title. In 3.6 these are arranged in order radiating from the center.

8.0 completely breaks the flow by sticking the address bar below the tabs. WHY. Also, where's my status bar, with all of my plugin status icons that I use constantly?

4.0+ is still broken, so I'm still going to use 3.6. Simple as that.

There's a difference between "broken" and "not the way I use it / am used to". Mozilla released a video explaining the transition of tabs [1]. Still don't like it? Right click somewhere on the title bar and deselect "Tabs on Top"

As for the status bar, just hit "ctrl+/".

The best thing about Firefox is that if there's something you don't like, there's probably an add-on for it. Try searching for add-ons that do what you need. [2]

1) http://blog.mozilla.com/faaborg/2010/06/24/why-tabs-are-on-t...

2) http://blog.fligtar.com/2011/01/16/how-to-customize-firefox-...

Why would I use 8.0 and spend time and effort turning it back into 3.6, when 3.6 already does everything I want or need it to?

Because it’s multiple times faster and being actively updated.

Address bar above tabs? Two clicks to change. Show status bar? Two more clicks.

I would love to use Firefox but I can't bring myself to do it. Firefox got really terrible and bloated and whenever I use it I am just waiting for a hint of terribleness to appear. Then I switch to another browser.

The main thing Firefox need to do is drop the version numbers all together. These numbers should be for developer's only. I could even argue developers do not need the version numbers and could work just fine with feature detection.

The fact is that most people do not care what version of Firefox they have. All they want is Firefox Latest. The latest and hopefully best version of Firefox will allow for the fastest and smoothest browsing experience. When I hear FF7/8/9 is out I sigh. Firefox automatically updates now, why all the grand announcements, especially considering for the majority of users the changes between 7 and 8 are fairly irrelevant background changes that won't be noticed.

Chrome is on what now? v15? Does Chrome continuously shout about how version 15 is out? No.. it just updates.. the user doesn't really know about it. The user doesn't really care about it.

Please.. just call the damn product [Firefox - no version number on the end] and be done with it.

First of all this is pretty much where Mozilla is heading towards anyways.

Firefox isn't bloated in terms of memory consumption, when compared to Chrome. It's interface has more nice features though, which I greatly enjoy though.

Did you actually test a recent version?

Oh btw. the Beta version of Firefox is an officially supported one, meaning if you find something that is bad for security bug you will receive the grant and stuff.

I have used a recent version of Firefox. It has a load of HTML5 goodness but not many website's run these bleeding edge features.

It still looks incredibly ugly on the Mac. The last time I used Firefox on Mac it was wobbly. I only now have it installed for testing. The change in versioning was a mistake which caused a load of confusion. Firefox may be getting better in performance or memory usage but does it matter? Other browsers are already there.

I will keep tabs on Firefox but I see no reason to switch from Safari or anything else for that matter to Firefox at the moment. It is a bit like IE tbh. It is getting better but for a generation of users the reputation damage is already done.

Reply on news.ycombinator.com