Slashdot Mirror
Firefox 3.0 Makes Leap Forward
Kurtz'sKompund writes "Mozilla has announced that Firefox 3.0 has passed a major milestone! The Places feature has been added to the alpha client slated for release next week. Places is a complete re-work of the bookmarking and history browser functions. It was at one point slated for Firefox 2.0, but will instead see release in Mozilla's next major version. '"We enabled the Places implementation of bookmarks on the trunk," said the Places team in a post to the Mozilla developer center blog. "Although there is still much to be done, this is an important milestone for us." Firefox 3.0 alpha 5 is scheduled to launch June 1. Because Places uses the open-source SQLite database engine to store and retrieve bookmarks and history entries, it's incompatible with earlier Firefox editions' bookmarks. Alpha users must convert their existing entries, Mozilla developers said."
Until they decide to remove it and delay it until Firefox 5 (after they skip 4)
I'm not the usual jerk that complains about FF's feature bloat, but I wonder if this new feature is optional.
Consider it a feature. They want to test all the memory in the machine for you ! Now that is what I call good folks.
All joking aside , isn't it on the bug tracker a lot ? I thought that the whole reason for this write was to fix the various memory issues before they moved on ?
This package Does Not Contain a Winner
Ok someone give me details on this "SQLite database engine" please.
I don't know anything about SQLLite but will this add any bloat/performance issues/etc the firefox 3.0? I mean a database engine to manage bookmarks seems like overkill to me.
The Mork file format was one of the most braindead database systems ever. Coding to access it is ridiculously difficult.
Mork is dead... thank the gods.
It's hard to leap when you're that big[1]. A few small steps, sure, but to qualify for a leap, I'd say one would have to replace HTTP or TCP/IP. And whether the direction is forward is also open for debate. If the browser takes up several times as much memory as the OS, I personally think it's time to find the axe and use it. Others may disagree -- which is why it's good there's more than one browser.
[1]: Obligatory "Look at the size of that thing!" quote.
Regards,
--
*Art
Look, I love firefox, but I can't really think of anything wrong with regular-ass bookmarks. I have no idea why they need to be in a database of any sort. This seems like a bloat feature to me.
However, the Mozilla SeaMonkey suite doesn't yet have SQLite. Will it be unable to share bookmarks with the new Firefox? Or will it get SQLite before Firefox 3 is released?
For anyone wondering what's going to change in FireFox 3.0 (Wikipedia quotes):
The largest known change for Firefox 3 is the implementation of Gecko 1.9, an updated layout engine. It will also provide CSS3 columns.[90] Firefox 3 will include features that were bumped from Firefox 2, such as the overhauled Places system for storing bookmarks and history in an SQLite backend, according to the wiki.
Also, what's expected to come in FireFox 4.0 (also Wikipedia):
On October 13, 2006, Brendan Eich, Mozilla's Chief Technology Officer, wrote about the plans for Mozilla 2.0, the platform on which Firefox 4.0 is likely to be based. These changes include improving and removing XPCOM APIs, switching to standard C++ features, just-in-time compilation with JavaScript 2 (known as the Tamarin project), and tool-time and runtime security checks.
Full Tilt
Hmmmmm, 285Mb with 2 windows and 2 tabs open. Only 18Mb shared too, which isn't a good sign for our multi user machines.
4331 me 15 0 285m 67m 18m S 1.7 3.4 0:27.10 firefox-bin
I don't know, is it windows guys developing it these days?
Deleted
I'm looking forward to this going gold for just one reason: some of the sites I visit frequently have a particularly in-your-face usage of auto-refresh which pisses me off (i.e. insisting on re-loading just when I'm in the middle of reading a particular paragraph). FF 3.0 (I heard) is supposed to be able to block this...
my bookmarks collection got so big that i removed them from my firefox profile and created several html pages (categorized) with only the links (without all the other meta info) and bookmarked the pages...
Politics is Treachery, Religion is Brainwashing
Well now it seams like they are starting to move into Microsofts style of software writing. To bad too I really liked firefox and it wasn't to bad on my solaris box.
I don't have enough bookmarks to need a db to track them.
This package Does Not Contain a Winner
All posts resembling the pattern “why don't they fix this problem instead!?” are off the mark, irrelevant, and just plain whiny. Just because some new feature is being added does not mean your pet peeve is going completely ignored. There is more than one person working on this thing, and as remarkable as it may seem, many software development tasks can be done in parallel! Imagine that: doing more than one thing at once on a project!
Why bother.
SQLite has this name because it is... lite! It has a very, very small footprint. By your comment, I can tell you have never used it. I have. It is designed to be small and easy to embed into an app with out requiring a bunch of resources.
General, you are listening to a machine! Do the world a favor and don't act like one.
In essence, sqlite is a file-only rdbms. No networking. No user permissions (apart from those of the host filesystems, of course). Works with mostly standard SQL. Very lightweight and easy to use (I've used it for a couple of projects of my own, so it can't be too hard).
On my Debian lenny systems, the sqlite executable is around 35k, and the library file around 150k. That's pretty small.
Backing a database up involves copying the file somewhere else. That's all. Add to that the fact that it's well-documented (unlike most previous FF and Thunderbird data files), and it'd be silly NOT to use it.
I love Firefox ( or Iceweasel as it is called on Debian ) but I am running a fairly streamlined xfce install and currently fire.. err... Iceweasel, is using about the same amount of memory as the rest of the system combined. I know there are other browsers, but I really do like Iceweasel, except for the memory footprint. Seeing that I only use a fairly small subset of the features it would be nice to have a light version with just the essentials. I wouldn't suggest axing the features other users love and depend on, but perhaps provide an alternative for those of us who really don't need an advanced database for our 3-4 bookmarks ?
PS: It would also be nice if Firefox didn't highlight "Iceweasel" as a typo.
I only just upgraded to 2.0! Sighs and reminds self that only dead products (and people) are static. We now return you to your normally uplifting /.
If I was deep this is would be profound, if smart then wise, if a poet then verse. Here it is, you judge!
Where are the times when it was a quick and lightweight browser I loved? Today... Konqueror > FF.
Ayi ayi ayi!
Former US House candidate, TN-5
Adding new features is not automatically a bad thing. It does not intrinsically slow down a program or make it cumbersome. Of course, these are two possible side-effects, but are not always certain. With good practices and architecture, new features are a boon, not a bust. Also, think of all the things the computer on your desktop does right now. Would you rather it have the functionality of a machine from a few decades ago because people complained that expanding its usefulness was counter-productive? Let products evolve, let engineers innovate, and let the process for coping with the consequences work.
I cannot believe some of the mundane topics Slashbots will harp on these days. Get over it and try adding some useful dialogue to the stories instead of bitching about things you do not understand or understand only as a result of experience with one particular vendor in Redmond.
Why bother.
As a user of multiple computers (work, home, friend's house), I use del.icio.us and the Firefox plug-in for it, and all my bookmarks are stored in a database that I can access from any computer. That's superior to this new "improvement". I think browser developers are really scraping the bottom of the barrel, looking in vain for "the next big thing". I'd rather see work done on useful plug-ins. That work well with existing browsers, than see a new browser that has some improvements of debatable worth that break the old way of doing things entirely.
Nothing to see here. Move along.
"Hmmmmm, 285Mb with 2 windows and 2 tabs open. Only 18Mb shared too, which isn't a good sign for our multi user machines"
.. 37 MB with 2 windows and 2 tabs open and any increase is to do with the size of the cache.
--
..
The alleged leak is to do with Firefox using memory to cache pages. The solution is simple. Type about:config in the address bar and schange browser.sessionhistory.max_total_viewer , browser.cache.memory.capacity and config.trim_on_minimize.
Firefox here
I like Firefox except for $RANDOM.FUD
was: How much memory does it consume? (Score: Firefox memory leak FUD)
davecb5620@gmail.com
Yes, everything should be done your way. Heaven forbid features other people find useful worm their way into your software, since you are obviously the anonymous arbiter of all that is technology.
Slashdot - where whining about luck is the new way to make the world you want.
This lightweight, fast, simple database eliminates many of the headaches associated with using a full-on SQL installation, and works just as well for most of what most developers and users need.
If you're a Perl geek, like me, you will find this Perl module for seamless SQLite interface to be a power tool. The next time you need to get something working by morning, and it's 2am and the person "in charge of databases" hasn't called back, you'll be thanking it.
technical writing / development
I felt the same before i tried tabs for a while, "I can't see anything wrong with just opening a new window for each new link i want to open separately".
I wouldn't judge it before i try it for a while.
So instead of having a nice simple bookmarks.html file it is all in some places.sqlite file. A few questions/observations.
1) If this is correct, disappointing that the devs called the bookmarks file places.sqlite instead of bookmarks.sqlite so people know where the bookmarks are if you want to move them. Am I correct?
2) Is this new file now no longer human readable the way bookmarks.html?
3) Someone please tell me how this makes my life better as a normal Firefox user? Sure sounds like change for change sake to me...
In the list of features slated for Firefox 3.0 is a sync api to allow for these kind of things. The places apis also allows for something called remote containers. Remote containers let you implement a folder with bookmarks from a different datasource, some examples that come to mind: file system, ie favorites, del.icio.us, etc...
I rather think that integrating bookmarks and history into a light database would be pretty decent. They can then put all my passwords, phone number, SSN, pictures of my penis, and home address in a database, and my life will soon be very Convenient(tm).
/.ers seem very reluctant to apply Reading for Comprehension to the summary, which clearly states a restructuring of the system. Maybe this new feature will make us all happy and the genocide will stop. Speaking pragmatically, probably not.
But really.
I for one welcome our query-driven overlords.
In case none of you have guessed, I'm visiting a friend who is having lady trouble, and I am therefore drinking. w00t!
Please stop stalking me, bro.
They're apparently embedding a fucking SQL DATABASE into Firefox 3. Given that SQL databases are not exactly known for being light-weight
There's a big difference between an embedded database like Berkely DB or SQLLite and an enterprise DB like DB2 or SQL Server... And embedded SQL Database is VERY lightweight.
09 F9 11 02 9D 74 E3 5B D8 41 56 C5 63 56 88 C0
The alleged leak is to do with Firefox using memory to cache pages. The solution is simple. Type about:config in the address bar and change browser.sessionhistory.max_total_viewer , browser.cache.memory.capacity and config.trim_on_minimize.
.. 37 MB with 2 windows and 2 tabs open and any increase is to do with the size of the cache.
--
..
Firefox here
I like Firefox except for $RANDOM.FUD
was; Re:Memory use (Score:1)
davecb5620@gmail.com
It's also worth noting that native form controls for Mac OS X were enabled yesterday, something Firefox's Mac users have been clamoring for since the 0.x days.
You think a SQL db that only takes up 250K is significant? Geez, what do you have like 4MB of memory? The SQLite DB is already in FF 2. So it will not add anything to FF 3. However, the bookmarks will now just take advantage of SQLite, which they currently do not. This will give you plenty of ways to sort your bookmarks and store more info for each bookmark. For example, in FF 2 you can give a bookmark a keyword. This way you can just type that keyword and go to that URL. I use can type "/." and press enter and get to /. with no clicks. I have a lot of keywords setup for my most used bookmarks. Bookmarks in FF are more than just list of strings. They have a URL, name, keyword and description. One big problem with bookmarks in FF 2 is the inability to sort properly. FF 3 should fix that now that FF 3 will be able to use Order By to sort how the user likes.
General, you are listening to a machine! Do the world a favor and don't act like one.
"Iceweasel, is using about the same amount of memory as the rest of the system combined"
.. 37 MB with 2 windows and 2 tabs open and any increase is to do with the size of the cache.
--
..
I wouldn't have known about the issue except that every time Firefox is mentioned on slashdot it gets talked about. The alleged leak is to do with Firefox using memory to cache pages. The solution is simple. Type about:config in the address bar and change browser.sessionhistory.max_total_viewer , browser.cache.memory.capacity and config.trim_on_minimize.
Firefox here
I love Firefox except for $RANDOM.FUD
was: Light version? (Score:2)
davecb5620@gmail.com
Firefox currently stores bookmarks in an XHTML file so it currently uses an XML parser to load bookmarks, I don't think you'll notice any negative performance hit from SQLite.
Bookmarks were strings 10 years ago, now they are multi-field records, especially things like live RSS bookmarks. The hole point of the "Places" component is to give more state and functionality to bookmarks and history.
I also think the SQLite engine is going to be used for the new client-side persistence framework which does need database capabilities, so it would be there regardless.
http://www.mhall119.com
No, but offloading it to SQLite means they can slowly remove the horrible parsing, storing, etc mechanisms for the current data formats. And it also means that, hopefully, tools will be developed for IT managers to easily sync bookmarks with their employee desktops, e.g, updating home pages, etc remotely.
How do you know it has memory leaks? Seriously, have you done a exhaustive memory profile? Opening a few webpages, closing them, and looking at MemUsage in the task manager before and after is hardly conclusive. Of course, I am assuming you are doing this, but if you have concrete data, please share. I can leave Firefox running for weeks averaging 50+ tabs at all times.
That's strange. I'm running Firefox 2.0.0.3 on a considerably slower box (single Athlon 1800, 1GB RAM), and scrolling speed is fine, whether dragging the scrollbar, using the mouse wheel, etc. I suppose it could be an OS-dependent bug (I'm running Linux, not BSD), or optimizations in the official binaries, or maybe a video driver issue?
I have never had my profile corrupted with Firefox. I use Firefox on WinXP, OS X and Linux and keep my profile in sync with Google Browser Sync. However, if your profile does get corrupted, the bookmarks file should be fine since SQLite supports standard RDBMS constructs to keep the file from being corrupted. Since SQLite uses a regular file (like Access, but much, much better), you can just copy it to where ever you want as a backup.
General, you are listening to a machine! Do the world a favor and don't act like one.
Firefox users want a browser that displays webpages. A browser that is fast. A browser that doesn't hog the whole computer's resources. A browser that never leak hundreds of megabytes after an hour of usage.
:(
Adding a whole new bookmarks system is nice, but does the user-base need it ? Or at least does it need it more than it needs a stable and fast browser ? I honnestly don't think so, and I'm sad seeing Firefox going farther and farther from it's initial goals as an Open Source project
____
nico
Nico-Live
"Yeah seeing as how I only have 5 tabs open (3 slashdot, 2 mostly plaintext sites) and Firefox has steadily increased its memory usiage from 170MB to 187MB now for me... Ooooh 188MB keeps going. 189MB. While I write this post"
..
Five tabs open, three slashdot, two plaintext, memory used less than 62 MB, while I write this post.
--
was: Re: How much memory does it consume? (Score:2)
davecb5620@gmail.com
Firefox currently stores bookmarks in an XHTML file so it currently uses an XML parser to load bookmarks, I don't think you'll notice any negative performance hit from SQLite.
Actually, it's a specially formatted HTML file, but it's almost XHTML so that's close enough.
But the big thing about it is that, being an HTML file, it's human readable. I can open it in Internet Explorer or Opera and import my bookmarks from it should I decide to stop using Firefox. I can open someone else's bookmarks.html file and copy out bookmarks should I decide to.
It's plain text, so it's human readable, which is great if for some reason the profile becomes unusable. (Most common time I need to do this is after a Windows reinstall, when I'm copying out old profile data from the old Windows install.)
You can also get bookmarks from a Windows profile into a Linux profile quite easily.
In short, bookmarks being plain text or in XML is a very good thing.
Replacing that with a binary blob is a very bad thing.
Thanks for the tip, but it doesn't work.
I installed that plugin and http://www.reuters.com/ still refreshes
sqlite is vary resistant to corruption. In fact, the only way you're likely to see it is if Firefox for some reason tries to touch the databases with its own code rather than sqlite. Further, a simple command line query will dump everything.
Even the phrase "launching a SQL database" indicates you're thinking of SQLite the wrong way.
A better way of thinking of it is this: Mozilla developers are removing thousands of lines of code with an unknown number of bugs for a simple data storage mechanism used in thousands of software products, including embedded systems. SQLite works. In fact, it works astonishingly well. We're gradually using it to replace most data storage in our own products.
You say "binary blob" like you won't be able to access the SQLite database without Firefox. SQLite is open source and public domain, you can download the programs needed to access the file on multiple platforms.
No, you can't open it in your text editor, but then again it's not a document it's a database. You can't get what you need for the Places feature from a document unless you subsequently treat it like a database, which would truly be a performance disaster.
But, we might be able to have our cake and eat it too. The Mozdev article says that your bookmarks.html file will be updated when changes are made to the SQLite database. I'm not sure if this will be the case in the final release or if it's just in place for alpha/beta testing.
http://www.mhall119.com
The thing I don't like about this change is that bookmarks.html is the ideal homepage. It's a web page that's stored locally, has no ads or extra junk, and has a list of links to the things that you personally want to get to regularly, updated automatically. The biggest thing I missed when I used Konqueror for a while was that it couldn't render its bookmark list in the browser window. Of course, it should be easy enough to have an extension generate a nice file from the bookmark database every time it changes.
Can you give a link of the site that shows slow scroll please? People can test if this has been fixed in 3.0.
There is a spark in every single flame bait point.
Well it might be light, but it performs like crap compared to MySQL. Compare Amarok with the built in SQLite db or with MySQL. Hands down, no question, it works much much better with MySQL.
Those of us who already have MySQL installed will want to use it. I hope they make it as easy to switch as Amarok does.
Give me Classic Slashdot or give me death!
Can you point to any evidence that indicates that it is any particular feature, or even a huge group of features, that causes this increase in memory?
I'm willing to bet that the reason Firefox uses so much memory is because it uses an absolutely huge amount of data. Bookmarks, history, cache, not to mention the bazillion images used by pages (both those loaded and those cached), plus the increasingly complex markup used on pages which requires a staggeringly huge DOM tree to represent, plus all of the layout code necessary to support the increasingly complex CSS specification...
Somehow I doubt that ripping out the existing bookmark code (a reduction in code size) and replacing it with a SQLite backend (which already exists, and is stupidly small to boot) is going to account for any kind of memory consumption increase compared to just loading up the latest version of Slashdot vs. the old version of Slashdot (as one example).
I hope it makes a small leap backwards. Ever since a couple of weeks before fully upgrading to Ubuntu 7.04, when I run Evolution 2.10.1 and Firefox 2.0.0.3, after a few hours (or maybe a lot of GUI and HTTP events), some combo memory leak fills my 512MB RAM and starts crazily swapping. I have to kill both apps and restart them, recovering their sessions.
Even if they just had watchdogs that could restart and recover session state, they'd be more useable.
--
make install -not war
Where's all the hearty IE bashing that I've come to know and love around here?
Off topic yes, but if you're using Opera then you can type /. into the bar and press enter, and get to slashdot.org.
Steve Ballmer? How's it going, buddy? I knew it was you when the second chair flew by!
"Only the small secrets need to be protected. The big ones are kept secret by public incredulity." - Marshall McLuhan
So export your bookmarks to HTML.
Microsoft plans on releasing its new feature, "Locations", with the next release candidate of Internet Explorer 7. It requires a small version of MS-SQL Server to be installed, which will be included, and requires 1GB ram and 3GB free hard disk space.
SQLite usually beats out MySQL and PostgreSQL for many things. However, it really depends on what you are using it for. SQLite was not made to be a full-blown RDMS. Here is a speed test.
General, you are listening to a machine! Do the world a favor and don't act like one.
Firefox is going to rapidly transform from a primarily agrarian economy dominated by peasant farmers into a modern, industrialized communist browser?
Slightly disreputable, albeit gregarious
I haven't done a memory profile, but I do have smth for you to try out. Go to this site and leave FF running overnight.
Well those numbers are old, and my experience suggests otherwise. But it could just be that the amarok devs did a shitty job of implementing SQLite. Or it could be that my collection is >60,000 tracks. Still this might be an issue for people with insane amounts of bookmarks.
Give me Classic Slashdot or give me death!
That site contains Flash. Have you tried it without Flash installed?
Do people still use browser bookmarks? I can't remember the last time I bookmarked something. I just google it. Or I start typing the URL in and Firefox completes it for me from my history. Or I remember where I saw a link to it.
Just about the only bookmarks I have are a set of tab bookmarks that I fire up when I start my browser so I get slashdot, gmail, our helpdesk and a couple of other things come up. Yeah, that needs an embedded SQL database...
Oh well, I'm sure some people will find it useful....
When are we going to be able to run multiple firefox instances in isolated processes?
Tyranny isn't the worst enemy of a democracy. Cynicism is.
Bookmarks, places, whatever. That's so last decade. . Ever since i started using the del.icio.us plugin, i've never looked back. Categorization, portability and mobility are essential. The thing that got me hooked is being able to share my bookmarks between computers, platforms and devices.
*shower*
Reuters uses javascript to refresh and refresh blocker only nukes the META tag type of refreshing. The combination of refresh blocker and noscript does the trick here.
"I regularly see Firefox go above 250 MB of memory, and nothing short of closing it entirely will get that memory back"
Firefox less than 60 MB, climbed just over 60 MB with five windows open and then dropped back witht a single window. Now just over 75 MB with ten windows open. Dropped back to just under 70 MB and holding and that's after stressing it by opening/closing multiple combinations of tabs and windows. If that's the memory leak issue then I guess I can live with it.
davecb5620@gmail.com
I don't install Flash in Firefox, but I routinely reach 200MB use by the end of the day with an average of about four tabs open; I've only had it running today for about 2.5 hours and it's at 135MB. I do have 23 add-ons installed, but many of them are low-usage (Add-N-Edit Cookies, BugMeNot, ColorZilla, Copy Plain Text, IdentitySelector, IETab, Location Navigator, MeasureIt, and SecureWire Web Relay [for SSL VPN]). Even those that get heavier usage (AIOS, DownThemAll!, Forecastfox Enhanced, Gmail Manager, Google Toolbar, Slashdotter) shouldn't be eating up gobs of memory. I routinely run two or three virtual machines at a time on this 2GB notebook, so Firefox isn't helping things there, even with swapping enabled. Even disabling the add-ons doesn't save me from enormous memory usage by the end of the day.
You can never go home again... but I guess you can shop there.
One thing that bugs me about software in general, and open-source in particular, is this constant need for developers to change things just for the sake of changing them. "This way is better!" they'll say, but really, it isn't. When is "good enough" just plain good enough? When I upgrade to the next release of Ubuntu, for example, I don't want to be forced to re-learn all the software I knew and develop new habits just because someone decided it would be cooler if it worked differently.
Case in point: GAIM/Pidgin. When I first switched from Windows to Linux on my desktop I switched from the "official" IM clients (ICQ and Yahoo) to GAIM. But I had to retrain myself to use control-enter to send an IM, rather than tab-space (tab to move my focus from the text-entry box to the "send" button, space to "click" the button without using the mouse). This key combination worked in all the official clients, but GAIM couldn't be configured to use it. I liked being able to embed newlines in my IM's using just the enter key, so I didn't want that to be my "send" key. So I retrained myself. Fine; I'm willing to make compromises.
But now the new GAIM/Pidgin comes along and, oh, hey, guess what? You can't configure it to use control-enter to send any more! No more embedding newlines with a simple enter key press! Nope, that's not the "right" way to do it! And remember that send button? The one that was so handy every time you just had to go to the mouse to teach the built-in spell-checker all those words or acronyms it didn't know? Well, that's gone, too! It, too, was the "wrong" way to do things. No, there's only one right way, and that way is to send using the bare enter key, embed newlines with control-enter, and never, ever click a "send" button. So now I have to retrain myself yet again. Thanks, guys.
So apparently that's what's happening with Firefox now, too. The concept of history and bookmarks, which is perfectly fine and has been since NCSA Mosaic, is now uncool. No, it needs to be replaced with something else. And if I don't like it I need to suck it up and just take the time to completely revise my work habits! 'Cause some basement-dwelling, self-appointed God of Computing said so, I guess.
I really don't mind change. I welcome it! But I want to be able to change on my terms, not someone else's. I took to tabbed browsing like a fish to water, if you'll pardon the cliche. But nobody forced it down my throat. And I'm getting pretty goddamned sick of developers forcing these things down our throats just 'cause it's the "next big thing." I'm a geek, too, and I love playing with computers just for the hell of it. But I also use computers as tools to accomplish other things, and I don't need my software getting in my goddamned way.
despite what you claim, it's not the cache
..
Cleared the cache and it's currently at 58 MB
davecb5620@gmail.com
Yes, this would be the best thing. I like to reuse my bookmarks for other things, up till now this could be done with some perl hacking, it would be nice if there could be a way to access it from outside now as well. This would introduce a security issue as well, though.
molmod.com - computing tips from a molecular modeling
What a fool believes, he sees, no wise man has the power to reason away.
Has firefox completely failed to meet it's original goal if I am contemplating a bribe to some developer friends to compile a stripped down version of firefox so it doesn't use so many damned system resources?
Left running for 24 hours, with google homepage and gmail up, I am looking at about 1GB of memory used.
This just does not seem reasonable to me, but I'm a minimalist.
Pretty Pictures!
What a fool believes, he sees, no wise man has the power to reason away.
"I can leave Firefox running for weeks averaging 50+ tabs at all times." ;-)
You DO know about bookmarks, right ?
kind of like the way gmail tags e-mails? You can have folders that are tags, and any bookmark that matches that tag will show up. when you create a bookmark, you can choose from a tree of all the tags you have. You organize tags into trees so that they display well from a menu. There probably already is a plug-in for that (right?), probably should just fold that into the main code...
There are many easy to use tools for working with SQLite files.
the worst example that comes to mind is the 'frys javascript nightmare/hell page'. aka, their online ads.
a dvid=32664
try this url:
http://newspaperads.mercurynews.com/ROP/ads.aspx?
takes forever to load (even on fast connection) and just really slows FF down noticeably.
hit page-up and page-down to see slow scrolling.
worse, use the thumbwheel in any 5btn mouse, give it a few spins and sit and wait until FF catches up.
--
"It is now safe to switch off your computer."
Have you also disabled all in-memory caching? Firefox keeps pages in browsing history in memory by default.
If you have NoScript installed, it leaks memory every time you close a window (no leak when closing a tab though). There are other examples of Extensions chewing memory up - try installing Leak Monitor to find out whether you have Javascript objects getting orphaned.
However, having NoScript installed seriously reduces the amount of Javascript running so it tends to be a net win on memory usage.
Cheers,
Toby Haynes
Anything I post is strictly my own thoughts and doesn't necessarily have anything to do with the opinions of IBM.
Or another long-standing bug: Firefox's apparent inability to print layered transparent PNGs, such as the type used for the route-line overlay on Google Maps.
Seriously, go try it. At least on OS X, you don't even need to print, you can just hit the "Preview" button from the Print dialog, when looking at a Google Maps "print view," and it'll show you the route-line-less (and therefore pretty damn useless) map that it's about to print. It does the same thing on any other page that uses overlaid transparent graphics in layers.
It's well known and extremely annoying, but apparently the FF devs don't care about printing bugs?
It's unfortunate, because it just makes FF seem very much 'unfinished' when, in order to print a map from one of the more popular sites on the Internet, you have to use an alternate browser. The same thing works just fine in Safari on Mac OS X, or IE on Windows.
They need to stop adding new features and fix some of the bugs that have been around for months or years.
"Ladies and gentlemen, my killbot features Lotus Notes and a machine gun. It is the finest available."
what you are missing is
1 the sqllite engine is already present
2 they are yanking out the older crusty Mork/Xml/Vhatever code
3 they are using this to simplify things and enable cool things like throwing your bookmarks online (obsoleting the 12 bazillion bookmark sync extensions)
Any person using FTFY or editing my postings agrees to a US$50.00 charge
Scrolling in firefox was painfully slow on my system. If i happened to go to a page that used a Japanese or Chinese font, it would grind my system to a halt. Disabling PANGO really helped. Start firefox with a script that does something like:
/usr/bin/iceweasel
#!/bin/sh
export MOZ_DISABLE_PANGO=1
Well I don't know about memory leaks but the Downloads dialog is a fucking disgrace to efficiency. Download 50 or so things and the entire programs slows very noticeably.
I remember a few years back when I first upgraded to a version of Firefox/Firebird/Phoenix (forget what it was called at the time) which used the Downloads dialog and instantly the whole thing slowed to a crawl. Why? Because I'd never cleared my downloads cache before and the brand new Downloads dialog had about 2000+ entries in it. I think simply opening the dialog took a couple of minutes and the Clean Up took about five minutes. All for what is essentially just a listbox! God knows what would happen if they ever tried to make the Downloads dialog useful by doing crazy things like telling you when a download failed to actually download anything.
OK, rant over. I like Firefox for the most part but that has really pissed me off for a long time. Glad I finally got that off my chest.
Spelling mistakes, grammatical errors, and stupid comments are intentional.
Or you could read the above message as "I'm not even feeling lucky enough to type 'sqlite database engine' into google and hit 'I'm feeling lucky', I'd much rather type 40 additional words into a textbox on slashdot.
It's quite informative that the Mozilla site displays the results of a micro benchmark to show their competitors in a bad light after their history of memory issues. Its a worthless comparrison and anyone with profiling experience can tell you how easy it is to produce skewed micro benchmarks. At this stage everyone is aware there have been memory issues with Firefox and one silly micro benchmark does nothing to change this.
Do not try to read the dupe, thats impossible. Instead, only try to realize the truth
What truth?
There is no dupe
I haven't bookmarked anything in years. I do have some old bookmarks (the youngest one is two years old, the second youngest almost four years), and unsurprisingly, more than two thirds of the links are dead. Anyway, I digress...
v er. For others I guess bookmarks are the better choice.
Are bookmarks useful? When I want to find something - anything - I use a search engine. Unless the page is really obscure (in which case a bookmark would have been useful, yes), I can find it in less time than it would take to browse through all the bookmarks, not to mention the time initially needed to bookmark/rename/tag/label/describe/nickname/whate
What I'd like to see in a browser is a history tracker. For example, I visited site A, then I clicked on its page links 1, 2 and 3, and eventually went from site A to site B to site C, on which I clicked page links 1 and 2, and 2 had a link back to site A (which I didn't follow). It would be neat to see a browsing diagram. In this case, a big circle named A would be connected with B, B would be connected with C, and C back with A. The links between A&B / B&C would be a solid (as I followed them) line with an arrowhead, but C&A would be a dashed light gray or something. The caption on the lines would be the link name.
I still don't have a ruby tag for furigana in firefox / gecko. Even MSIE6 has that... it's the only W3C standard it has over firefox.
You can also use "%s" in conjunction with keywords which makes them even better.
bookmark keywords
I use this all the time. For example "flix clerks" will do a netflix query for me. I use this even with google and have the search engine box disabled in the UI. Because the cursor location defaults to the url area when opening a new tab/window, I find this easier (you don't get the history of search terms like with the search box though, I'm not sure that's always a bad thing).
I see a word I don't know? double-click word, copy, cntl-t, d space, paste, enter - bam, it's very quick and I only use the mouse to select the word.
I can't remember.
I don't need to scroll at all on this page. Wait, you don't have Adblock, do you?
There is a spark in every single flame bait point.
The current implementation keeps bookmarks.html in sync with the sqlite bookmarks database. See Places bookmarks have landed! for details. Most notably from that post is this tidbit.
* Migration from bookmarks.html -> places.sqlite will happen only once. From then on, bookmarks.html will be kept up to date with changes in the db, but not the other way around. This means that after initial import, if you make any changes to bookmarks.html (say, by running Firefox 2 and editing your bookmarks) you will need to manually export them and re-import them into the bookmark-on-places nightly (see bug 381216). This means if you run Firefox 2 and Minefield or Gran Paradiso Alpha 5, when it is released, using a single profile Firefox will overwrite any changes. It is imperative for testers to know about how to use the profile manager.These people look deep within my soul and assign me a number based upon the order I joined. -Homer Simpson
There is no leak alleged. Simply an observation that it appears to be consuming 1/4 of the RAM of one of our X login servers.
/. open and adblock extension:
/usr/lib/firefox/firefox-bin
On this machine, with the settings you mentioned, with only
browser.cache.memory.capacity=0, browser.sessionhistory.max_total_viewers=0, config.trim_on_minimize=true
7642 1 54684 190536 7.8 12.1
That'd be 55Mb resident and 190Mb consumed in total. Firefox 2.0.3
The question is... Is it really using that much RAM or have the developers simply mmapped everything?
Deleted
What a fool believes, he sees, no wise man has the power to reason away.
No, you can't open it in your text editor...
Ah, emacs. http://www.emacswiki.org/cgi-bin/wiki/sql.el
Okay, maybe that's not quite what you meant, but it's close.
Places still writes the bookmark.html file.
It just doesn't read-it anymore (but you can tell it to import it back if you like)
BTW, sqlite stores everything in one file so nothing is really changed
if you wan't to save, you just have to copy one file
if you wan't to move/copy you profile, it's just a file to copy
it will be much more robust, powerfull and allow new things to be done.
also the sqllite code is stable and field tested by hundred of projects so it's a very good idea to reuse it instead of using some mozilla only solution.
"know of three or four projects clinging to SQLite despite various problems it causes, some of which are deemed features."
I would like to hear what projects these are. I have been considering SQLite for a project and would like to know any possible problems.
EA David Gardner -"... but the consumers have proven that actually what they want is fun."
I routinely have 50-100 tabs open. If I bookmark something, I'll just forget about it and never read it. Bookmarks to me are for filing away things that I have read at some point and found useful, not something that I look through to find something new to check out.
1) Are you saying sqllite is already in firefox 2.0? I haven't looked at the source code for firefox 2, though I use it, but it would be news to me. 2) I don't know what that is, I just hope the new code is faster / more reliable (or at least equally as reliable). 3) I'd be surprised if this ends up being simpler.
And the isn't memory freed/cleaned up when you close the page? Otherwise, by this single fact the only thing you can say is __that page__ has memory leaks.
(n/t)
Is a plain text editor one of them?
Using a database for bookmarks is just plain stupid.
of course it depends on how many sites you visit every day...
In Opera, I save my daily visited sites ( each in a separate tab) as a session, and open the browser - or any number of session sets that way.
I do beleive Firefox has a similar cpability now.
Opera alos now has "speed dial" (odd name - but it works) - set up 9 sites and upon opening a new tab - click on any one of the 9 sites presented and you are there.
Think for a second. You do realize that would defeat the user's purpose of visiting the page, right?
I've never used a web browser that isn't annoyingly unstable. IE 6 crashed reliably on certain sites (haven't tried 7 yet), Konqueror crashes with frustrating regularity, and Firefox is generally fairly decent by comparison.
huh? If sqlite is portable enough to be used in firefox, then why not thunderbird?
I wouldn't mind a gmail-like conversation layout with other features enabled by having a "lite" relational-DB handle all the mail/contacts and other stuff. Microformats anyone?
I want a browser! I don't want the browser to need a database! SQLite? This is too far!
Well, I guess a non-special enhancement deserves an un-special comment from someone who is apparently very "special".
/.
Man, I hate it when the idiots escape digg.com and come to
blah blah blah
No, that remark is just plain stupid. Stupid as in 'being lacking in information, knowledge or the ability to understand facts'.
:)
:)
But, you probably meant to say "Using a database for bookmarks would require me to learn something trivial about a different (and more functional) way of storing information so that I can continue to play with my toys the (arcane) way I am used to". Or something like that.
Seriously, I hear you. I've sometimes been glad that the Bookmarks were in text files. I could do 'stuff' to them. But, it sounds like there are still going to be ways to get to the information. Maybe slightly more involved than popping open vi or notepad, but very doable. And a small price to pay for functionality, performance, features, etc.
Try to not be a luddite, eh?
In Nature, stupidity is a capital offense. In human society, too many get off with less than a warning.
My histories tend to not be more than about ten pages or so per tab, and I frequently close out tabs as I finish with them, even if they were only used for one story. What I will do often on a site is Ctrl-click on multiple stories on a page and then read them in sequence, closing out each story as I finish with it. It's my belief that closing that page should result in all memory associated with that tab being freed, but when testing that by watching memory use as I close the tabs, there is little or no change in the overall memory usage, and that strikes me as something that should be addressed.
There may be a fix (I think I know what you mean with the in-memory caching), but I'm not going to do that because when in forums, I will often jump back several pages, and I don't need or want them reloading just going back or forward. Besides, if I have to look up a method to do it, then it's not something that I'm likely to want to be sending my parents through, and that's an important point, especially since they have much older, slower systems than do I.
You can never go home again... but I guess you can shop there.
Actually, it's a known issue. Comes up practically every time someone mentions Firefox on slashdot, and is widely discussed in other places as well, including b.m.o. The term "memory leaks" is technically a misidentification of the problem, but the observed behavior is almost identical to what you would see if there _were_ substantial memory leaks. The issue is well known and widely documented. What Firefox is actually doing is extensive in-memory cacheing, trading RAM for other things, like improved back-button performance. On systems with enough RAM to support it, this is a Good Thing(TM).
You can turn it off in about:config, but I forget the exact pref name.
Ideally I think Firefox should try to do a better job of automagically figuring out how much RAM it can use this way without creating performance problems, but that's not entirely trivial to accomplish. (It's not enough to know how much physical RAM the computer has, which I think is what Firefox currently does. It would also have to determine how much RAM other apps are using, among other things.)
Cut that out, or I will ship you to Norilsk in a box.
Ah, memory leaks when gmail is open. Now maybe we can discuss an actual, confirmed memory leak for a change. Of the six reported leaks with gmail, four are fixed in Firefox 2.0.0.3 and two seem to be Firefox 3 only. If you can still reproduce a memory leak with gmail up in Firefox 2.0.0.3, you should file a bug report to make sure the problem gets addressed. A browser using 1 GB of memory after a day of use certainly isn't reasonable. It sounds like a problem that should be fixed ASAP. Filing the bug report, including a set of steps to reproduce the problem, is the first step to getting it fixed.
What a fool believes, he sees, no wise man has the power to reason away.
SQLite is meant for smaller DBs. When you have >60,000 tracks you're going to be pushing the limits of 'small'.
Not only that, but when I SSH into my box from somewhere else, I'll no longer be able to retrieve a bookmark by simply peeking at the handy HTML file Firefox currently keeps them in. Will Mozilla be implementing some alternative plaintext bookmark file for such use, or will I have to go hunting for an extension to enable it?
Nobody else has this sig.
Firefox also uses XUL (an XML dialect) for its entire interface. Dozens of DOM trees has got to hurt.
"it's not about aptitude, it's the way you're viewed" - Galinda
I have a sqlite databases with over 1.5 million rows and it doesn't break a sweat even with complex queries. Perhaps the amarok folks don't know about indexes?
Done with slashdot, done with nerds, getting a life.
Can I just have Firefox use my existing MySQL server instead? Why does a browser need its own private SQL backend?
Maybe they'll take a hint from the wonderful folks making Amarok. Amarok lets you pick your SQL backend, and only forces SQLite when there's nothing else.
I guess you don't know anything about an embedded DB? SQLite makes "sacrafices" for the sake of low footprint/performance. For example, SQLite, treats all fields the same. Where a normal RDBMS would do checks to insure data type, SQLite, skips those things for speed/footprint.
Before you "cry wolf" over some technology being put into Firefox, why don't you read up on it? Hell, SQLite has been around for a long time. The feature-set is well known. Oh, and if you are using Firefox 2.x, well, guess what, SQLite, is a part of Firefox 2.x.
And you whole "point" about using a relational DB to store "flat data" is just silly. Firefox keeps a lot of info about a bookmark. That has to be stored somewhere.
Since SQLite is already a part of Firefox 2.x and since there have been no problems with the SQLite in Firefox 2.x, it seems safe to say that Firefox 3.x should keep SQLite and actually take advantage of the great SQL support.
I would love to be able to sort my Firefox bookmarks with no hassle. However, that is not currently the case.
Bitching about some technology that you do not understand is pretty chee-see IMO.
With that said
General, you are listening to a machine! Do the world a favor and don't act like one.
The extension SmartSearch was made for people like you. It basically ties together your contextual menu with the Quick Searches you've already defined.
Let's say, from the above example, you've defined "d" as a keyword to search Dictionary.com. With SmartSearch, you'd right-click on the word, and choose
[Search for 'turnip' on...] --> [Dictionary.com]
Best extension ever. You'll use it all the time.
- Alaska Jack
I thought this was in FF now?
I have 'Keymarks' in my bookmarks for cpan and a few other pages (and have since ff 1.5), so putting:
cpan DBD::DB2
in the address bar brings up the relevant cpan search.
Are there any new features that the Places feature offers? I imagine having a sql structure would be a benefit if you're talking about thousands of bookmarks, but it seems like overkill to me. Unnecessary complexity. (It could very well be I just haven't thought of what the developers are imagining it will be useful for).
Regardless of how "light" this is, it really sounds like feature creep to me.
Regardless of how you feel, sqlite ( or BDB) will be faster than trying to parse text files or xml files for the amount of config settings in Firefox.
It's a good idea.
09 F9 11 02 9D 74 E3 5B D8 41 56 C5 63 56 88 C0
Well, since firefox has the "Undo Close Tab" feature, I'm pretty sure that it doesn't release all the memory when you close a tab. I'm not sure when it does release the memory, or if it's possible to disable undo close tab so that memory will be cleared when closing a tab.
Anthropic principle: We see the universe the way it is because if it were different we would not be here to see it.
I know of plenty of projects using SQLITE. For megabyte to tens of megabyte size database, SQLITE is GREAT. For hundreds of megabytes, it can still be good, but needs careful design and tuning. For 100s of megabytes, use a client/server DB
We use it to back our web based products. The concurrency isn't great but can handle dozens of users easily. We have DBs ranging to the gigabyte size. To date, in the past 4 years, there have been virtually no serious issues with SQLITE.
Pros:
quality code, excellent free and paid support
public domain code
excellent SQL 92 compliancy, with the exception of GRANT/REVOKE, which aren't relevant on this kind of DB
compiles on tons of platforms from embedded to Windows/Mac/Linux.
Fast, especially for SELECTs
ACID
UTF8/UTF16 compatible
Full Text search module available (caveat: I haven't tried it and don't know how good it is)
Well documented
Thread safe
Here are the caveats:
- As noted, it's not client/server. So it's concurrency is limited. Multi users apps should carefully consider their usage of SQLITE. You can work around this up to a point of course, but eventually you'll hit contention.
- For 100MB+ databases, consider client/server also.
- SQLITE can only use one index (single or compound) in the WHERE column.
- INSERTS are slow unless you batch them. This is because you are opening, closing, and flushing each time. So if you do one INSERT at a time, you'll whine about the speed. The time to insert say 10 things one at a time, is in my experience about the time to insert 1000 rows when wrapped in a transaction
- Both pro and con: a db is a single file. That makes it easy to deal with, but can be problematic in some cases, especially larger DBs
Suppose I wanted to know when I bookmarked it. And how long it's been since I used it. And how many times. What if I want to have it expire in a month? And what if I want to make a few notes about it, categorize it, group it. Etc. etc. etc.
All those things are easy with a database, and mighty painful with a "plain text editor."
Out of curiosity, what video driver are you using for X? At least in Linux, I found FF scrolling (and any scrolling at all really) to be very slow when using the VESA drivers. Might be related if you're using something similar. Not really FF's fault.
Obviously you've never read about Mork. It would be difficult to come up with anything less portable than Mork.
I've heard Firefox will still create the HTML-like file, just not read it.
MoFo switching to an SQLite-driven Places bookmarks/history system 1.) Reduced final compiled codesize by a bit over 70KB (according to their own build system) 2.) Greatly improves data reliability 3.) Is either performance-neutral or in some cases actually improved performance And you guys are complaining? If that's bloat, sign me up!
Toby, care to tell me where you got this info about leaks?
A "leak on window closure" regression appeared in NoScript 1.1.4.6.070322 (an "unofficial" development build) and was fixed 3 days later.
Since then I keep Leak Monitor in my development profile, even if it's a pain because of Firebug's and Venkman's leaks... ;)
Later NoScript versions are completely leak-free (see the changelog).
There's a browser safer than Firefox, it is Firefox, with NoScript
I'll tell you what ticks me off about FF: the memory & processor usage in opening a window. This is something like 50Mb to open a new window. This can make it a couple of seconds to load. A work colleague decided to test whether this was the extensions (eg HTML validator) that increased the load. Nup, ~50Mb to open a window. He googled for details on how to fix this, followed the various instructions, still 50Mb to open a window.
My guess is that firefox runs a new instance of gecko and rendering tools for new windows. It doesn't seem to do the same for new tabs.
Note: this probably isn't a memory leak but it is pretty freaking annoying.
meh
Translation:
I don't like the way Firefox doesn't free all its memory when I close a tab. I know there is a fix to that, but that fix destroys functionality that I like associated with this behavior, so I won't use it.
Way to go, you fully understand why this is not a "leak" and you enjoy the functionality you get from it, but you are still complaining. Listen, closing a tab doesn't instantly give you back the memory because you might need to get back to that tab soon, so it keeps it cached, which is actually a very nice thing. If you don't like this, I'm pretty sure you can turn off the closed tab caching, I know the Tab Mix Plus extension lets you. So it might be a good idea to read around in those forums of yours and ask some good questions to figure out how best to make Firefox do what you think it "should" do.
As the example that's most burned into my brain, take the python jabber client.
The developers of the program, which was meant to be very lightweight, decided they needed SQLite to manage chatlogs. I mean, it's lightweight, right? And it gets such glowing reviews from EVERYONE. How could this go wrong? They put a good deal of work into setting this all up, and once they started making releases with SQLite various users started reporting problems.
Aside from the usability issues with people no longer being able to access their plaintext log files as they wanted to, many started finding basic performance of the client to be completely unacceptable. There were major memory use issues (I got to watch one computer thrash for ten seconds for every single message received) and, very irritatingly, a problem where the harddrive would be accessed every time anything was added to the database--a deal breaker for many laptop users.
This was all traced directly back to the use of SQLite, and it was determined that it wasn't a mistake on the part of the client's developers. The SQLite devs verified that it was SUPPOSED to act that way! They cited some weird, half-baked philosophical grounds for their design decisions.
Anyway, I left that Jabber client behind, as did many others, because the simple inclusion of SQLite had made it unusable. The devs had invested a lot of effort in moving to the SQLite backend, and last I heard they were still trying to dig their way out.
Take it for what it's worth and all, but I have NEVER seen an SQLite installation that actually improved the project. If you really need an SQL database, bite the bullet and insist on a real one. Just think hard ahead of time about whether you actually need SQL in the first place. Lots of projects seem to be applying SQLite as a solution to a problem that doesn't really exist.
Odd, because that's not been my experience of SQLite at all - I wonder exactly what the chat log stuff was doing, as that seems exactly the kind of thing that SQLite eats for breakfast.
When I read the SQLite mailing list regularly, I was amazed how often people used it for jobs I would have thought to be outside its remit/scope - logging was a common example. People would ask questions about SQLite and mention their multi-gigabyte SQLite DB files with millions of records and I would think "Wow, that must suck in terms of performance." When asked about performance, these people invariable say "Oh no, that's fine, that's not the problem I'm asking about."
So adding a record to a DB whenever a chat message comes in? Seems like gravy for SQLite. I'm curious as to what the hell could cause it to thrash for 10 seconds to perform a single INSERT. Maybe they had lots of indices configured? Even so, 10 seconds still seems like ages.
In my own experience I've written code that converted DBs to SQLite, and did, say, 1000 inserts in the process, and this took a fair amount of time (of the order of 30 seconds, which surprised me), but that's due to the locking model SQLite uses. Wrap it in a transaction and those inserts took 1-2 seconds, iirc.
I'm not doubting the evidence of your own eyes, but I do wonder what the heck was going on there.
Wouldn't that turn off the *whole* memory cache?
I was talking about the pref that just turns off the newer RAM-eating rendered page caching feature.
Cut that out, or I will ship you to Norilsk in a box.
What a fool believes, he sees, no wise man has the power to reason away.
When did I call it a memory leak? The only time that I mentioned a leak in this thread was when I was referring to another poster's mention of a leak in NoScript (which someone else said had been fixed).
I referred here to the way that Firefox handles memory, which may include leaks but also may include unnecessarily holding onto memory long after it's practically needed. There's little reason to hold onto the last ten tabs closed per window, plus their entire contents. Even if that's seen as a necessary feature, at that point it's best not to hold onto cached information, and it should simply be a list of links, which even with expanded information shouldn't hold cost more than a megabyte. Disabling holding onto that memory should not require an extension or an about:config change; I should be able to find it in the Options dialog box, which AFAICT does not have any such options.
I have no problem with a browser holding onto memory in the currently open tabs. That's behavior I expect in all browsers to a certain extent.
You can never go home again... but I guess you can shop there.
--This is what DISK CACHE is for... If you close a tab or a window, it should immediately free the memory. If the closed page is needed again, it can be retrieved from disk.
--Disk is cheap, RAM should be given back to the system whenever possible to conserve resources. Especially when you have the OOM-killer $disaster-abortion-abomination to worry about in Linux-land.
.
== WolfriderV6 == I'm willing to admit that *I just might* be wrong... Are you??
Seems fine here... Scrolling is smooth and the thumbwheel doesn't slow it down. What is your video card and which driver are you running? A crappy driver will slow down everything that scrolls text mixed with images, not just Firefox. For reference, I'm running a 32-bit Linux distro on an AMD 4200+ dual core, with 2gb RAM, a PCIE GeForce 6600 and NVidia's 9755 drivers. FF 2 runs fine (although it's eating 260mb RAM) with 26 tabs open. Come to think of it, I guess that's 10MB per tab...
GPL: Free as in will
I do like hearing that progress in being made on the FireFox browser, I use it quite a lot and the various extensions to same. Keep on keeping on, Guy
Guy Cook Internet Marketing and Consulting Solutions since 1995.
It's most likely memory related. Whenever Firefox leaks enough memory to start swapping on my machine (usually within a day of me starting it) choppy scrolling is always one of the first ways I notice.
It really annoys me that people keeps spreading that line. I, and many others, have done those config changes and still have to close Firefox regularly to reclaim memory. It takes a day or two at my regular usage for Firefox to eat 1-1.5GB of memory that it won't give back if I close tabs.
IIRC it was using transactions and everying. A number of python- and database-experienced people took cracks at the codebase, but nobody could find anything that was being done wrong.
The spinning up of laptop harddrives at every chat message was a huge deal to some users, so they searched through the SQLite access options to see if some buffering was disabled or something.
They found articles and messages from the SQLite developers talking about how SQLite intentionally flushes Linux's disk buffers and how this was absolutely the right way to do things. It was weird. Very weird.
As for the thrashing, it was as if some huge library was being unloaded and reloaded into memory at every access. Perhaps it was something to do with the python SQLite library, I don't know, but again nobody could figure out anything that was being done wrong.
Eventually people started hacking their private copies to yank that functionality out of there.
I have a daily cron job that downloads my bookmarks from del.icio.us. It's a one-line curl command.
"Those who consume the bulk of goods are those who make them. We must never forget this secret of our prosperity."
That's great. Good for you. Really. However, since you seem to have entirely skipped the first line of my post, I'll repeat it again for your benefit. I'll even emphasize it this time so you don't miss it.
Just because you don't see the problem doesn't mean that it doesn't exist for anyone else.
I don't doubt your observations. I know that plenty of people don't have any problems with memory usage on Firefox. But I also know that there are a lot of people do as well, and the attempts of either group of people to discount the other are equally ridiculous. Your experience (and mine) is not universal.
However, since it seems like the only thing that matters to you are useless and arbitrary numbers, my firefox.exe is currently at 125MB with three windows and 4 tabs open. Closing all of the windows except for the one that I am posting in brings it down to about 117. Firefox has been running for about 2 hours, and I have neither the patience or the desire to track how long or how many pages it took me to get to this point.
If I don't put anything here, will anyone recognize me anymore?
Easy on the adding of features. How about fixing Firefox to use automatic garbage collecting before throwing the responsibility of doing it at us? (read: keeping open tabs at a minimum)
Count yourself lucky. I've lost my bookmarks file several times -- and when FF goes down, it blanks the handy file on disk, too. I've scheduled a little script to copy my bookmarks elsewhere each night because of this. It has also happened to my wife twice, and a couple of friends several times. It hasn't happened to me lately, so maybe they fixed it, but it is (or was?) a real issue.
Slashdot quality declines as the number of hot grits posts decreases. - Provolt's Law, Apr-09-2005
Does this works for windows version too?
98 is better than 95 which is better than 3.11 and so on.
Does ME == 1000?
What about XP?