Mozilla 0.9.2 ReleasedFriday June 29th, 2001mozilla.org released milestone 0.9.2 today, which many at mozillaZine believe to be the best candidate for a 1.0 release yet, from an end user point of view. New items since 0.9.1 include 25 more top crash bugs fixed (as measured by Talkback) along with a new context sensitive help system, a new view source window, and a new preloader for windows called 'Turbo'. Following 0.9.2 are two releases, 0.9.2.1, which will be taken from Netscape's branch after they complete their release from it, and 0.9.3 which will be taken from a branch off the trunk, and will not be under the same drivers checkin control, as previously reported. ok off to try it out I have to say, I'm extremely pleased at the progress the mozilla crew is making. I only wish Capital One had the same appreciation - everytime I go to manage my account, with or without TLS, I get booted to the following page - https://service.capitalone.com/non_compliant.html Hope C.O. gets with the program - I'd like to keep using Mozilla. Mike I had an engineer come to me the other day to test a fix for his topcrash bug. It was printing to file on Linux when you were trying to use relative paths. I don't know about you guys - and yes, i use Linux as my main platform - but topcrash bugs being Linux specific? This browser *must* be getting pretty good when the topcrash bugs are on a platform with less than 5% of the market and probably less than 5% of the downloaded browsers. Joseph Elwell. i meant they ARE getting silly. :P Joseph Elwell Actually, Linux and Win32 are nearly even. I just checked some figures i had on hand for 0.9 and over a two week period, and it looks like linux downloads were about 80% of those for windows. It's kinda lame that this opinion persists. The Linux browser has a lot more traction than the Windows and browsers, because Linux has no credible alternative to Mozilla, while Windows and Mac do. And the Linux user community naturally attracted to open projects like Mozilla. Finally, Linux distributions are the only operating systems ever to ship with Mozilla. I don't see Microsoft and Apple shipping Mozilla in the near future. So, my point is that it kinda sucks to just come out and suggest that Linux is a small part of the Mozilla community, and that their topcrash bugs are somehow marginal. Ater all, there have been 175 Windows-specific topcrash bugs. I stopped using the builds off the Mozilla 0.9.2 branch yesterday, partly because I wanted to report some bugs that I'd recently noticed, but mainly because every time I downloaded a file, I had to go through a couple of dialogs to get the file to save to disk. I was at first reluctant, because I remember all too well the instability of the 0.9.1 nightly builds that resulted in the special stability period of 0.9.2. After more than a day of use, I have to say that the nightly trunk versions are better than the 0.9.2 branch. In the past few days, the problem with downloading files was fixed. I haven't had a crash except for a crash that I was experiencing with 0.9.1 and the 0.9.2 branch: trying to download Paint Shop Pro http://www.jasc.com/download_4.asp Go ahead -- just try to download it with Mozilla! Some other irritating bugs that remain are problems with arrow keys in textareas http://bugzilla.mozilla.org/show_bug.cgi?id=69740 junk appearing after tables http://bugzilla.mozilla.org/show_bug.cgi?id=86430 and problems with the download time remaining http://bugzilla.mozilla.org/show_bug.cgi?id=87327 So, I have to say that 0.9.2 is not good enough for 1.0 yet, but the trunk nightlies are rapidly improving! Have you search/filed a bug on the PSP download page problem? If you have please put me on the cc. It looks like a DOM problem, I like DOM problems ;). I haven't looked lately but I believe that JASC's website uses document.all and that is why their nice popup menus at the top of the page don't work and also why the forms screw up. Maybe not? Our Weblogic app works fine with IE and NS 4.7, but with Moz 0.9.2 we get: java.net.SocketException: Connection aborted by peer: socket write error at java.net.SocketOutputStream.socketWrite(Native Method) at java.net.SocketOutputStream.write(SocketOutputStream.java:83) Jun-29-01 15:20:04 ERROR DispatchServlet - DispatchServlet Error: java.net.SocketException: Connection aborted by peer: socket write error at weblogic.servlet.internal.ChunkUtils.writeChunks(ChunkUtils.java:114) at weblogic.servlet.internal.ServletOutputStreamImpl.flush(ServletOutputStreamImpl.java:121) at weblogic.servlet.internal.ServletOutputStreamImpl.flushWithCheck(ServletOutputStreamImpl.java:337) at weblogic.servlet.internal.ServletOutputStreamImpl.checkForFlush(ServletOutputStreamImpl.java:487) at weblogic.servlet.internal.ServletOutputStreamImpl.print(ServletOutputStreamImpl.java:191) at weblogic.servlet.jsp.JspWriterImpl.print(JspWriterImpl.java:136) at jsp_servlet._generic._jsp._jsppendtrans._jspService(_jsppendtrans.java:492) at weblogic.servlet.jsp.JspBase.service(JspBase.java:27) at weblogic.servlet.internal.ServletStubImpl.invokeServlet(ServletStubImpl.java:120) You are kidding I hope? That's a stacktrace from a server side app! If your server side app crashes (or in this case, one thread in the app dies) because of something a client does, then it's the server side app that sucks - not the client. >That's a stacktrace from a server side app! Yes, of course it is. But the fact remains the browser is broken. This app works with IE, NS 4.7, and previous builds of Mozilla. Netscape 6.1pr1, I can't say, since it doesn't even handle session cookies correctly - I can't tell you how disappointed I was about that. If I get more time on Monday I will try to debug it. Perhaps others here with J2EE web applications or highly dynamic websites can share their results. And not only is it a stacktrace from a server side app but it's a stacktrace from the actual server - I don't see the actual app in there anywhere. Seems like WebLogic (or the one thread that handles that one connection) just crashes when it's trying to send some data to the client and the client isn't there like it assumed it would be. To get an exception thrown from a situation like that is just wrong and *definitely* WebLogic's bug and not Mozilla's IMHO (ok, maybe not so humble. ;) Where are (s)RPMs? 0.9.1 release did include them! Why 0.9.2 does not? Any special reason? The person who makes Mozilla RPMs is out of the office for a couple of days. Check back in a few days. Be patient, someone is likely to roll a few RPMs in the not so distant future. Remember, the "big 3" generic versions first... Mozilla 0.9.2 changes changes focus for the user? I'm viewing browser window A and click on a link to goto another URL. I switch to window B to do some reading while window A (now in back) loads. I get switched to window A when Mozilla's ready for me to look at it? Am I wrong about this? Is it a feature? tom It is called a bug. Mozilla has many of them. Mozilla 0.9.2 has fewer serious bugs than the previous Milestone (hundreds fewer, feel free to query Bugzilla for the exact number). This particular bug is bug 77675 http://bugzilla.mozilla.org/show_bug.cgi?id=77675 (please do not file this bug since it is already known. please do not comment in this bug unless you have some new _technical_ information to add to the bug. thanks.) --Asa ASA, this *bug* is part of the operating system. It affects IE 4.x & 5.x as well. I doubt that this one can be fixed. :( Mybe I've missed this somewhere in the discussions that were going recently following Mozilla milstone releases, but I'd like to know whether the news reader will have colapsing and expanding topics. Is this plannet to be included in v1.0 or it has been left for later due to some relative complexity. Thank you, Peter (mar22@usa.net) It's had that for a few builds now, and since 0.9.1 or thereabouts it seems to be rock solid. (The newsreader was in sad shape for awhile, but recent builds have made tremendous improvements. However, I digress). To turn on threaded reading, view the newsgroup and then choose View/Messages/Threaded from the menu. You can then expand and collapse topics by clicking on the arrows displayed. Cool that worked, but if I hit the Date after View Threaded, it "unthreads" to show me the messages by date-- really I want it to do both..threaded with the top thread listed by date (or alternately, the most recent addition to the thread would throw it to the top) any solution? W Yeah, they changed this a couple of milestones ago. It used to be that you could turn threading on and off independently of the sort method (by Subject, by Date, etc.). Now, "Thread by Date" is its own sort method. Threaded mode is always by date. And like the other modes, you can switch between ascending and descending order by clicking on the sort mode a second time. Personally, I prefer the old way. Sometimes I *like* to thread by Subject, or by Size. There's probably a line in prefs.js that I could change, but I'll be damned if I can find it. 7 seconds is too long on a p3 600mhz. This should be the top concern before 1.0 My start speed is even slower. The speed in which a closed window clears from the screen when multiple windows are open is around 3-5 seconds. Opening multiple windows gets progressively slower also. I run a 333MHz PII with Win2000Pro. I am still loving Mozilla, though. Previous builds took 12 seconds on my ThinkPad PII 128MB Ram, Win2k! That's unbearably slow when browsing with multiple windows and especially painful when IE5.5 on the same machine closes in under a second. Something is seriously broken here. http://bugzilla.mozilla.org/show_bug.cgi?id=87113 I was using Mozilla on an NT machine at work through yesterday (no longer...I don't work there anymore) and closing time was never noticeable. However, shutting down right after closing Mozilla sometimes gave me a "this application is not responding" alert for NSPR::EventReceiver. Is that what you mean? I personally think that the loading speed isn't the worst performance problem. After all, you do it once at login, you move Moz to a virtual desktop and you switch to it when you need Internet access. So it's not really important... For what *really* matters is the huge amount of memory Mozilla uses: 35M is a lot too much for a browser. Netscape 4.7 took only 12M. I don't see any reason why Moz should use 3x times more memory than Netscape. On my Win2000 that now have been up for four days without closing any browser app, Mozilla (2001062704 - using as primary browser) takes 26,4 MB and Netscape (using it for mail) 13,3 MB. 26,4 MB after four days without closing. I think that is quite ok. are you a windows user? Windows people seem obsessed with browser load time, for some reason. seven seconds isn't an eternity you know! You could watch half of an average commercial! If you _are_ under windows, try running mozilla -turbo when windows loads. That way you can have _two_ browsers that load on boot! I just installed 0.9.2 on my PIII 600 and it loaded in less than 3 seconds without the turbo option. Just loaded it onto PIII 750: 2-3 secs without turbo option... <1 sec with turbo option... wow! This will kick some [insert your name of browser here] rear end. Try loading Dreamweaver and Photoshop, shutting them down, and then loading Mozilla. Tried opening photoshop, etc. and startup on mozilla is still <1 sec with turbo and <3 sec without turbo. How much RAM do you have? I think this is one of most important factors. I have noticed that when I got additional RAM (from 128MB to 384 :)) startup time reduced significantly I have a PIII 450 and Mandrake 8.0. Doubling my RAM from 128Mb to 256 made a huge difference in start time. 0.9.2 is still noticeably slower than other apps but it is quite useable. With the Jun 13 nightly I can login to the bank site, but with the 0.9.2 release it expires my session and logs out the session immediately. I have tried changing the SSL 2/3/TLS options, has something else changed since the 13th with SSL? Some of these problems are caused by banks themselves rejecting Mozilla clients, but let's find out. Please file a bug against the PSM product in Bugzilla: http://bugzilla.mozilla.org/ Please include all the information we need to reproduce the bug, including the URL to the bank, what you clicked or typed, and what other things you noticed that might be clues. Also, please indicate if you see the same behavior on a fresh profile. Thanks, -Bob I have the same problem. The CIBC (https://www.pcbanking.cibc.com/english/servlet/SignOn) site has always worked for me, up to and including 0.9.1. With 0.9.2 it no longer works. When submitting my login information (bank card number and password) I'm presented with a page saying, "This page cannot be viewed with the method you've chosen." This is also an SSL bank site. (And, no, I'm NOT going to publicly reveal my bank card number and password such that this can be reproduced! <grin>) Jason. 0.9.2 SSL works fine for me with the Royal Bank of Canada (www.royalbank.ca) and VanCity Savings Credit Union (www.vancity.com). My wife banks at Royal Bank and you're right, it does work fine there. Just not at CIBC. I don't know what they're doing differently. Jason. FYI, I just located a Bugzilla entry for this problem - so anybody being annoyed by it can rest a bit easier knowing it's being looked at: http://bugzilla.mozilla.org/show_bug.cgi?id=87440 Jason. Mozilla (latest nightly builds) has a terrible time with Livelink 9. Livelink 9 instituted small icons that, when clicked on, display drop-down menus for many of the functions that required drill down techniques in Livelink 8. Since Livelink is a very expensive knowledge base web application ($125K per server), I think it behooves Mozilla to support it. Problem is, it's an account based system, so it's not accessible for those that don't have accounts to open access servers. Part B: I can help with debug info if someone is interested. I don't know how to decipher the Java/Javascript Consoles nor do I have time. I would rather download a talkback nightly, see what doesn't work, and report it to an interested party. Yeah, I could submit a bug report, but then again, how would one test it? PS: I can't use IRC at work (it's blocked). Any ideas... I would argue that it is Livelink's responsibility to support the browser, but that is beside the point... I went to Livelink's home page and took the guided tour. I assume the broken menus you are talking about the java applets on the side bar. I've noticed that a lot also, Mozilla tells me to go download the latest Java Plugin, but being a Java developer I already have the latest version! I think this is a known issue with a workaround, anyone? Personally Java Applets for dropdown menus is overkill to me, I personally programmed a similar application, Project and document management, forums.., to Livelink and used Heirmenus to the same end (and much more effectively imo ;-)) recently. It works very well in Mozilla... Damn, $125K? I should package up my app ;-) jon Yes, one could argue Opentext should support the browser. And then our experience is that IE 5.x doesn't do some things properly that Netscape 4.7x can do properly. Opentext support will tell you to use Netscape. The Newsticker is a Java App (bottom left) and I had to grab the NPJava130_01x.dll's from the Netscape plugins directory plus the NPOJI600.dll too. This make the Newsticker sort of work. It seems to not be quite right. Yet in the Opentext support site, it took the newsticker a very long time to load, but then worked okay. I discovered this when I would try to "Get the Plugin" and Mozilla would never see it. It installs the plugin dll's in the Netscape directory only and I guess Mozilla doesn't see them there. If you go to the what's new in Livelink 9 link, you'll see a screen shot of an actual Livelink display. The "popup" menus at the upper right don't work (just goes back to the front page) and the "one-click" function menus don't work either (same thing). Lots of Javascript of which I can't decipher. This is the type of Javascript Console error I get. Error: Unable to convert JavaScript value NaN to Java value of type int Source File: https://knowledge.opentext.com/knowledgesupport/functionmenu.js Line: 87 I can email the functionmenu.js file if anyone's interested. from the functionmenu.js file I can see a 'NS4' and 'IE4' which suggest that they are using browser detection. Nothing is mentioned about Moz, so I presume it is running as NS4? Or maybe its failing because it can't decide whether the browser is NS4 or IE4? Mozilla's useragent field can have that effect on badly written javascripts. are you familiar with bug 86103? http://bugzilla.mozilla.org/show_bug.cgi?id=86103 If not take a look. You might also want to see if any of the bugs listed in the link below is related. http://bugzilla.mozilla.org/buglist.cgi?bug_status=UNCONFIRMED&bug_status=NEW&bug_status=ASSIGNED&bug_status=REOPENED&email1=&emailtype1=substring&emailassigned_to1=1&email2=&emailtype2=substring&emailreporter2=1&bugidtype=include&bug_id=&changedin=&votes=&chfieldfrom=&chfieldto=Now&chfieldvalue=&component=Live+Connect&short_desc=&short_desc_type=allwordssubstr&long_desc=&long_desc_type=substring&bug_file_loc=&bug_file_loc_type=substring&status_whiteboard=&status_whiteboard_type=substring&keywords=&keywords_type=anywords&field0-0-0=noop&type0-0-0=noop&value0-0-0=&cmdtype=doit&namedcmd=All+My+Bugs&newqueryname=&order=Reuse+same+sort+as+last+time I'm certainly not one to stress general performance improvements -- I'm much more concerned with standards compliance and stability -- but the incredibly platform discrepancy in performance is astonishing. A page such as http://overosa.dhs.org/musik.html loads ridiculously faster on Mac OS 9.1 (300 MHz G3, 192 MB RAM) than on my Linux/X11 box (600 MHz K7, 256 MB RAM). And it's not even worth mentioning Windows. What is to be accounted for this great difference in performance? I'm guessing it has something to do with the amount of suck present in linux/X11 specifically X11 being so high. The other two operating system's level of suck compared to X is much less. thank you The layers for Linux are: Mozilla->Gecko->GDK->X->hardware But this shouldn't affect performance visibly. As you can see, Mozilla must call Gecko to do something, Gecko calls GDK, GDK calls X, and X calls the hardware. It only needs 5 functions. But calling functions are VERY fast! If you write a function that calls 5 other function you will not notice any slowdown. When I use a GTK+ program, it draws just as fast or faster than a Windows program or an SVGAlib program. So your assumption that the slowness is caused by the layers is completely incorrect. Not so fast. X11 is a network based protocol, in principle there can be a couple of context switches involved in sending a graphics command to the X server and getting the response (Yes, I know commands can get buffered, it depends a lot on whether the call requires a round trip or not). There can be something of an art to getting good performance out of X since it requires getting rid of as many round trips as possible. If the X display is local, it is possible to write an Xlib that goes through some kind of shared memory area or a pipe rather than a TCP network connection. It would also be possible to write an Xlib that could eliminate the context switches altogether if it detected that the display was local, and write directly to the screen buffer, though depending on the display architecture this is likely to be unsafe. Some X implementations do have optimizations of this type, but not all do; we'd have to know the versions of Xlib and the X server in use on the machine in question. Also I don't know if Mozilla on Linux is linked statically with Xlib or with a shared library - this would limit what optimizations an X server could do. Bruce I'm not only seeing strange performance discrepancies among different platforms but even between similarily configured Windows and Linux (don't know about Mac or Solaris) boxes even with the same hardware and only a little different OS version (like Win95 / Win98). I really can't figure out what the problem is, but it seems quite widespread. You can often read things like "it's so slow on my machine XXX" and and answer saying "I've got also XXX and its fast as hell". Weird. I can not even get the page to load at all with NS 3,4,6 moz 0.9.2 or IE 5.5 or 6.0. What kind of example page it that? The speed of Mozilla for Windows running in Wine (yes it does sort of work!) compared to Mozilla for Linux running on the same machine (at the same time) is also astonishing. The only thing I can think of is that more optimalisation work has been put into the Windows code. Actually this is a problem with 3 Alpha PNGs layed over. This is probably better optimized on Moz for Windows and Mac. I will report it as a bug. I just noticed that in order to get LDAP Auto-complete working on my Mac with 0.9.2, I had to delete my existing directory preferences (that worked well with 0.9.1), create a new directory entry with the same information, then quit and restart Mozilla. Is this a bug or a feature? Brian! I really can't figure it out. I want to use Mozilla, I like Mozilla, there are no bugs that totally ruin the experience for me, so why the hell am I still using Opera 5.11? Maybe its that it takes over 8 seconds to load on my P2-400 192megs system. Maybe its that when using the super cool preloader, I lose a ton of memory. I tried to play a game with the preloader active and found that my framerates dropped to an unplayable level. Maybe its because Mozilla doesn't have Opera's super cool gesture system http://bugzilla.mozilla.org/show_bug.cgi?id=76537 . Maybe its because I've set Opera as my default browser and I'm too lazy to change it. I really want to use Mozilla. I really do like it. But there is something about Opera that keeps me coming back. I really wish I knew why. Everytime I download a nightly and discover some cool new feature, or a bug thats been fixed I say to myself that I'm going to use Mozilla now for everything (including baking frozen pizza's). It just never happens. I think the biggest reason is the lack of gestures though. Its so easy to go back a page or two in Opera, whearas in Mozilla I have to move my mouse all the way up to the top left of the screen. Anyone know if any progress is being made in that department? Note: I'm not trying to flame anyone working on Mozilla. They have done tremendous work and I thank them all for it. hmm when i want to go back I either right click and click on back or alt left arrow I don't know of any other app that uses mouse gestures "I don't know of any other app that uses mouse gestures" Pretty much just Opera, and Black And White. That's it AFAIK. It seems a little gimmicky to me. I don't use most of the Opera gestures, only the forewards/backwards ones and I have to say that once you've used them there's just no going back. The speed is just fantastic (of course the speed with which Opera pulls pages out of history is a factor) and it is incredibly frustrating not to have it. Of course Opera has its downsides (mostly the DOM as far as I'm concerned) but I think that their ability to think differently gives them a nice competative angle. in Mozilla click your right mouse button and move the cursor to the lower right and let go. Some say you've just selected the first entry of the context menu (back) but why not believe it was something like a mouse gesture? :) Of course you have to be used to it to hit it every time but I think it's fast enough In Mozilla the context menu triggers onclick on windows and onmousedown on unix I think. So on windows going back (especially 2 or three pages) is more troublesome than on MozLinux and significantly so compared to Opera. I don't expect to see gestures on Mozilla simply because of the community nature of the project. It's the nature of the beast that it's difficult to make unilateral design decisions like Opera have made with gestures. Mozilla still rocks though. I think before we get into anything with mouse gestures we should make sure the regular UI is solid. That means full keyboard accessibility (I just filed a bug to get F1=help in mozilla #88739), help everywhere, perfect cut/paste/drag and drop, clear dialogs, etc. Then we can think about mouse gestures. Otherwise we'll have a mess of a UI just like Opera. It is tough to use plug-ins in Opera, yes/no? I find that it is, otherwise it is great. They are doing a great job! But the plug-in problem, phew... This is the best one yet. "Following 0.9.2 are two releases, 0.9.2.1, which will be taken from Netscape's branch after they complete their release from it, and..." Does anyone know if Netscape will be taking any of the more critical fixes from the trunk before their final release on the 0.9.2 branch? I'm thinking specifically of some of the XPCDOM regressions which are just now being fixed. These leave a pretty big hole in our DOM standard compliance and functionality, and I'd hate to see NS6.1 released with such ugly regressions. I would agree. There have been a few crashes too introduced by the XPCDOM landing. There might still be more that are undiscovered or not reported. ok this isn't really related but it's so cool I have to post it here. Let's say you want to have several websites as your home page. For example http://my.netscape.com http://www.mozilla.org and http://www.mozillazine.org So in the location for homepage prefs put in javascript:window.open("http://my.netscape.com","_content");window.open("http://www.mozillazine.org","_blank");window.open("http://www.mozilla.org","_blank"); When you startup mozilla those 3 websites will load. You could also make it a bookmark (and give it a keyword) so instead of opening several related websites individually, you load them all at once.
Would it be possible to add something like a sidebar tab or just a button to set your homepage with all open pages at this time? Or add something like this to the closing code (exit is just a JavaScript what I know) Operas sessions just in Mozilla :) I believe you could simulate this multiple homepage scenario using a bookmarklet as your home page. A bookmarklet is a javascript: URL. In this case it would open the three windows. You could also create a custom home page that does this using JavaScript embedded in the page. That is a bookmarklet... What happens when you create a new window though? I would have thought it would open up 3 new windows. I use Sweetmail email client for Macs and 0.9.2. I just tried to copy my member validation code from Sweetmail to the page in MozillaZine to validate my group membership and found <paste> greyed out. Shouldn't I be able to copy from one program and paste to Mozilla? What happened ? - You have to restart Mozilla when changing theme. - The preferences are badly imported from a NS4.x installation (proxy ports are set to 0!). - The toolbar (modern theme) is not as nice as before. Yea, it sucks, preference changing on the fly has been broken for a while. Suppositivly it doesn't work right, though it always worked fine for me. So they decided to turn it off, untill they can get it fixed. :( It takes a while to get used to the modern3. But if it seems really weird your profile might have sucked the modern theme from netscape6.1pr1 which doesn't load right in new builds of mozilla. If so try creating a new profile from scratch. "The toolbar (modern theme) is not as nice as before." This has been fixed in the nightlies. See bug 88576 http://bugzilla.mozilla.org/show_bug.cgi?id=88576 for details. Alex This page http://www.mozilla.org/projects/security/components/configPolicy.html tells you how to block pop up windows, prevent web pages from resizing browser windows, moving browser windows, and finding the your screen resolution and color settings and more What they need in Mozilla is a checkbox on the front of the UI (like on the menubar) that lets you set or unset whether pop up windows are enabled.... I know this is kind of off topic, and Im not sure if anybody else has posted on this, cuz I just hopped on here really quick, but I am a beta tester for CompuServe and I just got the new beta of CS 2000 7.0....anyways it comes with MOZILLA as the default browser!!! Now, I dont know if you are all as excited about this as I am, but it might just mean that AOL could include Moz as their default browser, which could be the push it needs to become the dominant browser....!!! AOL need to do something fast. Checkout the stats on Webreference.com for browser usage. In case they change, they looked like this at the time of this post. Yesterday 26980 Visitors OP 0.26% NS 3.34% IE 88.60% 6 months ago NS's share was 15%. I asked Andy king at Webreference.com if the NS figure included Gecko/Mozilla and he said it did... If this keeps up don't be surprised if AOL pull the plug on Mozilla. I thought AOL had nothing to do with Mozilla? Besides, MSFT will end up not renewing its contract w/AOL, and Netscape will be shipped w/the next AOL, and immediately, IE will lose 30%+ of that share. AOL owns Netscape and is funding mozilla, partially to sell its technology to embeders (yes they are free to use it if they give their source, but many embeders are willing to pay for closed source), and hopefully to be used as the new browser for AOL 7.0 Relativize statistics, especially those at webreference.com. I'm following these very tightly too and noticed that there are some very weird user agents like IE55 (fifty-five), Netscape 8, 9 or put your preferred number in here: ___. Also seen some funny Opera 99 lately. Of course these are small percentages of users, but all those that changed their UserAgent string from Opera to MSIE cannot be discovered. (yes, the same old story I know ;-) ) Opera's user agent string spoofing IE actually still contains the word "opera" in them. "but all those that changed their UserAgent string from Opera to MSIE cannot be discovered" Yeah, that must be like.. twelve people at LEAST.. Maybe fourteen! I might be alone in this, but I set my preferences in Opera 5.12 as "Identify as Mozilla 5.0". (Not a bad idea if you want to give Netscape's statistics a little boost. #126 Re: Re: Re: Relativize! (was:Re: AOL needs to do sby garfieldbond Wednesday July 4th, 2001 9:56 AM Technically I suppose Opera is closer in terms of tech to Moz than in MSIE. If opera spoofs as MSIE, someone is always trying to push me an ActiveX control. *sigh* Three percent seems a bit low and probably represents the fact that web developers (webreference's natural audience) tend to use IE much more than NN due to superior DHTML capabilities in IE. thecounter.com shows June 2001 NN share at 6%, IE at 86%. For June 2000 the shares were 16% NN, 78% IE; for December 2000, NN 10% and IE 84%. The NN falloff has been especially sharp since March, while Mozilla-based browsers have not yet reached a significant fraction of one percent. Strauss: I was very surprized (or maybe not) that today's statistics are something like 3% for Netscape, while it has been around 11% for the last 2 weeks. Dropping in one day from 11 to 3 is a little bit too hard to believe. Notice that MSIE (all) had around 88% in the last 2 weeks as well, so what's the point? Webreference decided that some netscape UAStrings don't belong to Netscape. Silly. Really. I guess we would have to know whether they used to be overcounting or whether they are now undercounting. Do you have any specifics that would resolve that question? In any case, statistics gathered from a single web site are unlikely to be all that useful. It's quite possible that users of one web site are atypical. That's why I look at thecounter.com instead. Its statistics are gotten from many sites. If you compare its free figures http://thecounter.com/stats/2001/January/browser.html to paid figures from a professional research house like WebSideStory http://news.cnet.com/news/0-1005-200-4928051.html , they're very similar. I think the worst thing about these stats are what is above Mozilla in the listings. Here is the current list: 1. MSIE 5.x 254691049(72%) 2. MSIE 4.x 42613129(12%) 3. Netscape 4.x 35488809(10%) 4. Unknown 6558431(1%) 5. Netscape comp. 5693823(1%) 6. MSIE 2.x 1944480(0%) 7. Netscape 3.x 1183842(0%) 8. MSIE 3.x 1100529(0%) 9. Mozilla 5.x 992747(0%) 10. Opera x.x 663642(0%) There first 3 are pretty much what I was expecting, and I will ignore 'unknown' or 'netscape compatable' browsers cause I don't know what these are. Now look at items 6-8: IE2, Netscape 3 and IE3. These are ancient browsers which can barely handle frames, let alone things like CSS or the DOM. IE2 didn't have much of a marketing push, and was terrible, so why is it above Mozilla 5.0? Even bearing in mind that Netscape 6 had only been out for a couple of months when these figures were taken it is pretty amazing. -- After writing the above, I found later statistics. By May 2001 http://thecounter.com/stats/2001/May/browser.html Mozilla had around 0.27% market share, which is still lower than other reports I have heard. This had risen to about 0.29% in the June report, which puts it below Opera. I'll see if I get get any better stats out of this. OK, I have done some graphs, and have the data in an excel spreadsheet too. Take a look at http://www.lemnet.com/browser_stats.xls (which includes 3 graphs). If you want them in other formats, you can contact me on: ICQ: 8895922, Yahoo/AOL: thelem3, MSN/Email: ian@lemnet.com I could be wrong but I believe that win95b shipped with IE 2.0. You'd be supprised by the number of offices still using win95b who have never upgraded their browser. As far as I'm aware, Windows 95 OSR2 (Win95B) shipped with IE3, not IE2. I had the misfortune to use IE2 once when my ISP inexplicitly switched from Netscape to IE. Within hours, I'd changed back. Alex I think that's incorrect.. Win95b shipped with IE2.. IE3 didn't arrive until much later and was a completely different animal (MUCH improved with even some limited CSS support). IE2 was *VERY* primitive.. I'd say about on par with Netscape 1.1N Yeah, I still use Windows 95 on one of my computers (and have the CD). It comes with IE 3. Today at work I discovered there was an NT machine that still had IE 2 on it... heh heh (cringe). Big difference. oh, that's Win 95b by the way. Win95c came with IE4. It might be possible that there is a version of windows 95b that was transitional and still had IE2 I suppose There was a Win95C? I always thought the bundling went like this: Windows 95 - (No bundled browser) Windows NT 4 - (No bundled browser) Windows 95 OSR2 - Internet Explorer 3 Windows 98 - Internet Explorer 4 Windows 98 Second Edition - Internet Explorer 5 Windows 2000 - Internet Explorer 5 Windows Millennium Edition - Internet Explorer 5.5 Windows XP - Internet Explorer 6 Alex Here are composite stats from 50+ sites in Europe. The stats are collected yesterday and are damn near identical to the stats over the past couple of weeks so they should be very reliable. The sites vary a lot in topics so the stats should be pretty balanced and unbiased. Draw your own conclusions: MSIE 5.x - 69.38% Netscape 4.x - 14.75% MSIE 4.x - 14.32% MSIE 6.x - 0.70% MSIE 3.x - 0.29% Netscape 5.x - 0.22% Netscape 3.x - 0.19% other - 0.10% Netscape 6.x - 0.02% MSIE 2.x - 0.01% IBrowse - 0.00% Netscape 2.x - 0.00% And today, they're showing Netscape, MSIE, and Opera with 0.00% each. It tends to do weird things. It'll sit reliably at 11%-13% for a few months, then drop to 2% for a few days, then go back to normal. Of course, what do you expect from WebReference? Read a few JavaScript articles and you'll get some idea of just how little technical knowledge those guys actually have. As long as Mozilla development continues to focus on the sexy parts (optimization, for example) and less on delivering a polished product with stable APIs, we're not going to see high rates of user adoption or companies other than Netscape basing products on it. If they depend on the use of off-site cookies, then the figures will likely be too low for Mozilla. Most mozilla users are probably (at least for now) pretty web savvy users, and will set the browser to reject off-site cookies. Some of their other figures look dubious as well. They show about 1.5 million Linux users, whereas according to counter.li.org the figure is closer to 17 or 18 million. Another thing I noticed was that the figure for Win 3.x actually went UP from June 2000 to June 2001. I suppose this is possible if people with really old machines just went out and bought modems, but it still seems rather odd to me. can we call 0.9.2.1 the ip address release? ;) ..back to random crashes again :( does anyone know what's causing the sudden instability? I haven't noticed this on Win98. I hope you use a talback enabled build :). Personally, I don't use talkbacks for the last 2 weeks because installers with talkback aren't still available (since June 18) and I hate using .zip builds. Hoping they will fix this build problem soon. I haven't seen any random crashes using the latest nightlies. Maybe you got a bad build -- get a new Talkback build and see if it still crashes. the past 3 days think i had probloems with the nightlies, but now everything is back to normal with the current batch :) Mozilla has not crashed for me in the last few weeks except when I try to download Paint Shop Pro http://www.jasc.com/download_4.asp This is in spite of downloading the latest nightlies as they become available. The rock-solid stability that people were talking about a few months ago really is starting to materialize! I think I'll try leaving Mozilla running for days at a time so I can find some more crashers for the team to work on. Everytime I start 0.92, it starts in a different position. It does not seem to remember the last position it was in when I closed it. I like it in the upper left hand corner. I did not have this problem with 0.91 and earlier. And because of this, the tiling of multiple windows is inconsistant. And if I resize a window for a certain website, it then seems to make that the default window size, even if I close the window before opening another. And when I restart 0.92, it uses the last resize even if it is not the last window to be closed. Three steps forward, one step back is still progress. :-) Thanks Mozilla!!! I'm really dismayed by the progress of the product, and continually amazed by the optimism some of you guys still have. I'm still using Navigator (4.77) but I've given up hope on seeing a great IE-killer any time soon. Mozilla is still way too slow even though my computer's not much slower than the average out there. I see crashes that live on from a few months ago (such as the Preference-change crash). And what's up with those unprofessional cosmetic bugs like deformed dialog boxes and unreadable dialog text? If something this simple still can't be fixed, what can? Maybe I'm cynical but I never thought AOL wanted to do much for Netscape's browser. (One last comment: I see how people say concentrate on the browser and forget about the mail client, etc. Yet I still use the email software in 4.x because it's so good. It's not too much to ask for a browser suite, not just a barebone. The development community can't deliver though, I think largely due to resource constraints. Anyway, this is symptomatic of the whole project. It's doing so poorly that we continuously limit our ambitions.) I don't know, mozilla is faster to render pages for me than IE is by a good 5 seconds on anandtech.com and my computer is a big piece of crap. A k6-2 450 with 192 megs of ram, and an s3 virge. I think too many people had way too high of expectations for mozilla and are getting dissapointed, me I've been impressed by every new release since moz .7 and I've been using it since M12. I've only stopped using it as my main browser since .7 for a month because of some really heinous focus issues that have gotten fixed since then. For what I do, and how I browse it's already the best browser out there. Far better than IE and Opera and it's been better than netscape 4.7x for months. That's just my opinion though. I've never gotten it to crash on pref changes though. Maybe the developers are having a hard time reproducing that one. Peace out, happy 4th of july, time for me to get drunk. I think too many people had way too high of expectations for mozilla and are getting dissapointed, me I've been impressed by every new release since moz .7 and I've been using it since M12. ------ Yeah, that expectation is IE. Mozilla renders pages at about the same speed to me, and that doesn't really matter all that much. Is IE really rendering pages all the slowly? I've never had an issue with it. Startup times are much slower. One can point out that IE is embedded in the OS, but it's a moot point when you have a end-users who know nothing about how the system works start up a browser and tap their foot. Something more transparent the -turbo is going to be needed if the world's largest user-base is going to accept this new browser. I've been impressed by every new release, too, **when compared to previous releases**. IE still feels like a much more solid browser. And furthermore, why can't Mozilla release a more Windows-friendly chrome? One of the reasons I use IE is because it fits in with my consistent graphical interface. Buttons look like they should. Scrollbars don't feel different. I'm not switching UI contexts when I open my browser window. When I use Mozilla, I feel like I have a foreign application running. That's not what you want users to think when they run your application. Mozilla does not feel like a Windows application and as such, doesn't seem like one should be running it on a Windows system. Pretty GUIs are nice, and the new Modern interface is just okay (personally, I hate the fact that it's that depressing dingy blue, but that's me), but consistency is more important to the simple user. Why do you think Microsoft strives so hard to make its interfaces consistent across its product lines? They should deliver a Mac chrome with the Mac build, a Windows chrome (which ties in with the current set Windows color/font/size settings) for Windows builds, and the default chrome for the Linux builds. Having filed insane amounts of Swing Look & Feel bugs in Java and being really anal about [platform] consistency, I'm surprised by your comments. In general, I think Mozilla's Windows L&F works amazingly well and looks amazingly like the real thing. 1) hmm heard of Quick Launch? 2) IE and other windows apps are grey and completely lifeless-how depressing can that be? I want windows apps to be consistent with mozilla -not the other way around. Yeah, but when I change the colors/fonts/sizes of my Windows (which I can do from the Display control panel), every one of my Windows apps changes with it .. except for Mozilla. I use a pretty warm grey for my Windows, so it's not cold at all, especially when compared to the cold dingy grey that Modern uses right now. I want my Windows apps to behave like Windows apps. Themeability is great, and certainly Modern is looking cleaner all the time, but the point is the end-user *needs* consistency, and defaulting the distribution to start with an interface that looks *nothing* like the two standard interfaces the majority of end-users (who know nothing about themes) are used to using means you're breaking the consistency and implanting the idea that Mozilla is a "foreign" application, not a "standard" or "real" one. hmm I changed the menu color from the display panel and it changed the color of the menu bar in all of my Windows apps (including Opera) except for mozilla and IE and Windows Explorer... So are you using classic or modern? classic look like ns4.x and uses your system colors. Modern doesn't for obvious reasons. "Yeah, but when I change the colors/fonts/sizes of my Windows (which I can do from the Display control panel), every one of my Windows apps changes with it .. except for Mozilla." Hmm, that's strange. My mozilla changed font size and color just fine when I adjusted it in the Windows control panel. I even tested some of Microsofts accessibility schemes and the high contrast oversized black and white scheme was picked up just fine by Mozilla. I'm guessing that you're using Modern which doesn't pick up system colors and fonts. Try Classic. It seems to work fine for me. "and defaulting the distribution to start with an interface that looks *nothing* like the two standard interfaces" Um, not sure what Mozilla build you're using but about 499 of the last 500 Mozilla builds that I've downloaded default to Classic which has an interface that looks a lot like the win9x,NTx standard and behaves as most users would expect it to on those platforms including picking up the system colors and fonts. --Asa If you're seeing "cold dingy grey" there must be something wrong with your monitor. Every copy of Modern I've seen has used a rather pleasant pastel blue colour. Maybe your monitor is broken ? Umm, Mozilla comes with Classic as default, in case you haven't noticed. And with Netscape 6.x, themes are one of the touted features. I'm quite sure the end user will know about it when the day is over. I *like* the look and feel of Modern. It feels like the windows app I never had :) And on mac, it certainly feels like the mac app I deserve. what pref change crash bug? The only thing I can think of is that mozilla used to crash when changing a color but that's fixed Well I'd like to say a few things. First of all I really don't mind at all that mozilla looks that different to native windows applications. Secondly if you want a more classic look just switch back to the classic theme, no big issue. Plus so far i've never heard anyone complain about winamp not having a native windows UI. ;-) Thirdly I don't think this project is performing poorly. How good was netscape 1.0 or should we say mosaic 1.0 plus it was no way near as complex as mozilla now is. Remember near all the code had to be rewritten. Of course it will lack some features but it is still going to be version 1.0. I'm sure it will make more progress after that. Don`t get me wrong i don't think we should rest on our laurels but I think there has been a lot of effort put into mozilla and a lot of progress made. I find it quite unfair for the people developing mozilla to get some much bashing ;-o As spend lot of my time under Linux OS, I am looking closely at mozilla since a very long period (downloaded and tried each milestone since M12). But I must confess I still can't see Mozilla as an IE-killer application : when I talk about mozilla with friends, only few points remain : - Mozilla is [slow]/[bloated]/[...] Mozilla's performance has been recently improved, but when I look at Donner's graphs (http://www.mozilla.org/mailnews/win_performance_results.html), I think that most of performance work is already done, and that we won't reach those goals (we may see a 20% improvement, but not a 100% one !). Mozilla is really enjoyable if have a big CPU and enough RAM (I can live with that - that's what I have), but I don't think it will become an IE-killer for low-end configurations. Like C++ => Java, the IE => Mozilla move could be an effect of Moore's law ? - Mozilla JS is not compatible with IE nor Netscape 4.X JS. They say : why bother rewriting JS for 1% of the browser market ? What are opinions on this point ? Creating an IE-compatible JS layer (how could AOL use Moz without such a layer) ? Hoping that the massive adoption of mozilla will make web-designers updating their HTML code befor moz-users get tired of such issues ? JS is a load of crap no matter what browser you're talking about. Both Netscape and Microsoft did the right thing and deferred to the W3 on markup languages like HTML and XML. Why they've both obstinately refused to do so with JavaScript/JScript/whatever (ECMAScript? Yeah, right. Don't make me laugh. Netscape knows that was a smokescreen move) is beyond me. Until there's a rigorous standard defined, JS is never going to work right. I doubt you know what you are talking about here. First off, I wouldn't compare JS and HTML or XML since I believe they are different script/language. (VBScript would be a closer comparision.) As for your statement about standard defined? This may surprise you... IT IS... Actually, you mentioned it, ECMA is a standard defining organization. I believe the origional message is more referring to DHTML related suite of script and language such as CSS (which BTW Mozilla has better support) instead of simply the problem of JavaScript. "Both Netscape and Microsoft did the right thing and deferred to the W3 on markup languages like HTML and XML. Why they've both obstinately refused to do so with JavaScript/JScript/whatever (ECMAScript? Yeah, right. Don't make me laugh. Netscape knows that was a smokescreen move) is beyond me. Until there's a rigorous standard defined, JS is never going to work right." A rigorous standard? ECMAScript *is* a rigorous standard for the JavaScript syntax, and both IE and Mozilla follow it pretty closely. Or maybe you mean the Document Object Model, which defines the API available to embedded scripts? There's a rigorous standard for that too, from the W3C. It's the W3C Standard DOM (http://www.w3.org/DOM/DOMTR) level 1 (DOM1) and level 2 (DOM2). What you're failing to grasp here is that it comes down to a choice between supporting IE and supporting IE+Mozilla+Opera+Konqueror. Since the same amount of effort is involved either way, the latter is the clear winner. In any event, pages that fail to function due to lack of IE4/NN4 JavaScript support are almost nonexistent. I was pleasantly surprised to see that web designers in general are less incompetent than I'd previously believed. >>it comes down to a choice between supporting IE and supporting IE+Mozilla+Opera+Konqueror. Since the same amount of effort is involved either way, the latter is the clear winner. << Excuse me? Have you ever been involved with a professional QA effort? QA cost is directly proportional to the size of the test matrix. Even if all the browsers were identical in their behavior and there was no development cost (in fact they're all very different and impose major development costs through these differences, but let's ignore that for now) you are talking about a 400% increase in the QA cost to cover the expanded test matrix. >>pages that fail to function due to lack of IE4/NN4 JavaScript support are almost nonexistent.<< That's because it costs so much to develop and QA DHTML code that most development houses punt on any serious client-side interactivity and keep as close to a basic HTML feature set as possible. You've inverted the causality -- the compatibility situation in the installed base of deployed web pages is good because the JavaScript/DOM/cross-platform/cross-browser compatibility nightmare has prevented people from doing much of significance using DHTML. So why hasn't a pretty major layout bug such as this one (http://bugzilla.mozilla.org/show_bug.cgi?id=65816) been worked on since April? I would think that those sorts of things would take some sort of higher priority. Can't say I agree that that is a "pretty major layout bug" when compared with other layout bugs. Of course it would be nice to see it being worked on but there are other bugs I'd rather see developer and QA resources working on. --Asa This bug is one of those "this site I like to visit doesn't work! I don't care if their markup is crappy, Moz has to support it!" Nonsense. There's absolutely nothing wrong with the markup in any of the test cases and the bug has been around since January. It's a simple case of the layout not being handled properly. I take an interest because it affects me personally, but I'm not the one that originally reported it, and I've seen three more sites in recent weeks that have been affected by it, which is what drew my attention back to it. Obviously, when people are making valid sites that work in other major browsers like Opera and IE, it's not a "Moz has to support crappy markup," issue. It's a "Moz isn't fully supporting valid markup," issue. As more and more sites move to these sorts of CSS/P-based sites, the browser which is supposed to be the most compliant ever had better learn to support them. Ignoring it and blaming it on "crappy markup" won't make it go away. That kind of attitude just leads to a product that everyone dismisses as a wannabe contender to IE. I reported this bug <a href="http://bugzilla.mozilla.org/show_bug.cgi?id=55694">http://bugzilla.mozilla.org/show_bug.cgi?id=55694</a> over 9 months ago and it has not had any attention at all. I appreciate there are more important bugs but CSS-P is important and becoming more so. My company actually use this bug to justify not moving to any Gecko based browser as it totally f**ks up about 20-30 of our internal Intranet sites that use CSS-P layers, which works perfectly in IE5+, NS4.08+ and Mozilla up to something like M10. http://bugzilla.mozilla.org/show_bug.cgi?id=55694 TonyG, you should have read the text above the reply form. Mozillazine doesn't allow HTML. well, i finally decided to build mozilla on my low end machine (which took a hefty 20 hours) but.. it build with debug messages :( as mentioned on the building mozilla page on mozilla.org, i downloaded the source and did a 'make -f client.mk'. knowing very little about compiling i assumed this would do the trick, but.. i ended up with a very slow mozilla build. so, what arguments need to be passed to disable debug messages, enable various features etc? i'm sure i'm doing something wrong.. I suggest you take a look at this http://webtools.mozilla.org/build/config.cgi basically --disable-debug will do the job You want to add the following to your .mozconfig file: ac_add_options --disable-debug ac_add_options --enable-optimize If you test Linux or Mac OS X against the standard of "Windows-killer," they fail miserably. If you test Dr. Pepper or Orange Crush against the standard of "Coca-Cola-killer," they fail miserably. Yet Linux and OS X, Dr. Pepper and Orange Crush are all successful products. I don't understand this obsession with "market-share" and "IE compatibility." The marketplace is huge! And as for that 1% augment... If the company you work for can afford to ignore 1% of its market, then your company is too small to be of any significance. If the multi-national I work for ignored 1% of its market, then a whole hell-of-a-lot of us would be out of a job! 1% of next to nothing = next to nothing! 1% of a hell-of-a-lot = a hell-of-a-lot! "I don't understand this obsession with "market-share" and "IE compatibility." The marketplace is huge!" Have you not seen what IE's been doing with it's 85% market share? They can control the future of the Web, determining which features get included or excluded and which technologies get adopted. They already tried to do that with Smart Tags. They do it all the time with CSS "enhancements". If someone else can't get enough market share (I would propose that 33% is a reasonable start), IE will become the de facto standard (not HTML or CSS or DOM, but IE) and Mozilla and other browsers will have no choice but to support IE. All sites will be written for IE, and anything that doesn't comply will be seen as useless. 1% of a hell-of-a-lot is a hell-of-a-lot. 99% of a hell-of-a-lot is a hell-of-a-lot-more. Try getting a job in a design firm and explaining to your boss why you should spend the extra time and money to support Mozilla when 99% of your customers use IE. It costs a lot less to convert 1% of the people to IE than it does to convert 100% of your code to Mozilla/IE compatibility (hypothetically speaking; if IE were to become that dominant, you'd presumably see a shift in that browser away from standards compliance). In that case, IE would go away from the current standards but there WOULD be a standard. It would just be Bill Gate's standard.. "Try getting a job in a design firm and explaining to your boss why you should spend the extra time and money to support Mozilla when 99% of your customers use IE. It costs a lot less to convert 1% of the people to IE than it does to convert 100% of your code to Mozilla/IE compatibility" I can't see IE ever having 99% marketshare. Supporting Mozilla and IE5+ with shared code is quite possible and reasonable to do. We already write two code paths (Netscape4 versus IE), and it makes sense to merge IE5 and Mozilla, especially since it's so much easier. The argument is that Mozilla is the next version of Netscape. We're used to writing special code for the different browsers. If you can show that you can write basically the same code for multiple browsers, that's a big win. Convert customers to a different browser? There's no way that we could convert some of our customers to using IE. They like Netscape quite nicely, thank you very much. Some of them choose Netscape because of the better security. Some because they're using alternative or older Windows OSes. It remains to be seen what they'll do now that the 4.x line is dying. Netscape 6.0 was terrible (and really didn't support what was necessary in the corporate environment). I expect they'll move to Mozilla as it approaches 1.0 and Netscape 6.1 is released. "I can't see IE ever having 99% marketshare. Supporting Mozilla and IE5+ with shared code is quite possible and reasonable to do." I can see them having 99% market share. On Windows, it's almost exclusively the browser of choice now as its integrated into the system. Someone has to specifically choose not to use IE to use something else, rather than choose between IE and another browser. Macintosh is fast approaching that as Netscape's relative instability on that and Microsoft's quick work in getting IE working under OS X has begun locking up that platform as well. The only place IE doesn't even have a foothold is Unix-based systems (like Linux, Solaris, etc.), and the market share for those systems on desktops (which is where the majority of web browsers are) is decidedly tiny and with the release of OS X, the impending release of WinXP, and the failure of Linux companies and organizations to make a truly consistent and usable Linux desktop that works for the end-user, looks to be hitting a wall in acceptance. So, when we get back to the main operating systems that have 97-98% of the desktop market (WinXX and MacOS), you see IE being given not chosen. To make people switch to another browser, they need to have a good reason (speed benefits, memory benefits, better UI enhancements, more stability, etc. etc.) to even switch to it. If that doesn't happen, it's only a matter of time. Right now, more and more people are moving away from coding for Netscape4 and IE5. Part of it is the WaSP push that tells people to abandon that in favor of standards-compliant pages. Well, IE and Mozilla are really the only two major browsers with the ability to support such pages (Opera's brand-name strength is negligible, but possibly growing). When you factor in that Mozilla isn't finished yet, you have people more likely to switch to IE over Mozilla. IE's market share on some sites I've been visiting is something like 88%. That's almost 9 out of 10 browsers. That's insane. Soon, "standards compliance" will mean "IE compliance", which no one in this community really wants. How will that happen? Simple, Microsoft will introduce very easy-to-use and slick tags that do very useful and inventive things. Page designers will easily pick them up for sites because 90-95% of surfers use a browser that supports them. Those stuck with Mozilla are either not going to get the full-experience or not going to get any experience. That's why market share is important for Mozilla. Mozilla is fighting an open-standards battle. Supporting Mozilla, though, is only worthwhile if Mozilla can garner enough marketshare to make it more cost-effective to support both Mozilla and IE vs. code for only IE and lose business (ie. the amount of money you lose from having your designers test and code workarounds to make things work in both browsers is less than the money you lose from lost sales to Mozilla users). Even at 90%, I'd be willing to bet that those two are relatively equal, especially when you consider a cheap design department can do a site in FrontPage (which will produce IE-compliant code) a pittance. I have just got back from Microsft TechEd 2001. I went to a session entitled "Building powerful DHTML solutions with IE6". They gave a potted history of the browser wars and made some statements about IE6's standards compliance, including that it is 100% Dom level 1 and 100% CSS1 compliant. They then went on to demo loads of nice stuff like sortable tables, vertical text, bidi text, horizontal tabs and about 10 other groovy thangs. All these things used MS Behaviours to function and are thus absent from any Gecko based browser. I love Mozilla but I tell you, MS will get 90% market share by end of this year unless Mozilla/Netscape 6.x makes some sort of a modest comeback. Commercial developers are already coding to the emerging de facto standard. This standard is dead simple. If it works in IE then its fine. Budget, time, effort used to even test in Netscape/Mozilla - never mind modify code - is increasingly hard to justify. Sad but true. Again - as far as commercial web development goes - to create a standards compliant page you code it for IE. At this moment in time Microsoft are paying lipservice to W3C standards. They may even be sincere in their commitments to these standards. But I ask you, if they turned around in the morning and stuck 2 fingers up at W3C, how many developers would stop coding for IE? 1% is a lot in simple mathematics but effort to payback ratios mean its a simple choice to make. If the cost of satisfying 1% of your users is more than the potential payback - then you wring your hands, apologise profusely and then say sorry - your not worth it. Happens all the time. Bottom line - Mozilla and Netscape need to take the erosion of market share seriously. I personally think Mozilla is just about ready. A bit more UI cleanup and polishing. some more topcrashers sorted out. Memory footprint and performance optimisation (taking account of Moore's law) and then its ready for the last major bit of development. The last major round of development is proper evangelism. I look here to Netscape not Mozilla. Mozilla advocacy is impeccable. Asa and others posts on /. are just one of many testimonies to this. Netscape doesnt have a proper site for the product. the link on netscape.com to the browser is buried and bloody nonen gives a damn. You used to read Eric Krock or Mitchell Baker etc on OReilly or suchlike.. now you get silence. AOL Time Warner - biggest Multimedia company on this planet - biggest ISP, biggest IM provider. A houisehold name. Do you think these guys know how to do advertising? Damn right they do. so where is the Netscape marketing? hopefully its awaiting a viable product and the Marketing giant is poised to make the whole world think Netscape is cool again. you ever seen the UK adverts for AOL with the girl in the Cyber dress? I would use Mosaic if she told me to!!! <End Rant> "you ever seen the UK adverts for AOL with the girl in the Cyber dress? I would use Mosaic if she told me to!!!" OK, you're strange. I find the girl (she's called Connie) rather irritating. The fact that she's transparent is rather concerning too. <random factoid> Apparently, the father of the woman who plays her is a Liberal Democrat MP. </random factoid> Alex Connie is a sight more appealing than that of anyone one else I can think of in the entire computer world. She is only irritating in the amount of times per day she pops up on my TV But that does kinda prove my point in that noone in the UK with a TV does not know what/who AOL is. Thats brand penetration and thats what Nwetscape needs. You might like to know that the AOL "Connie" ads won an award recently...for being the most irritating TV ad campaign shown in the UK in the last year. Apparently, Connie is seen as patronising and irritating in equal measures by the people surveyed. I'd be surprised if we ever saw Netscape TV ads in the UK (they launched their ISP "Netscape Online" in the UK first with no obvious ad campaign), although we have had Yahoo!, Excite and Lycos TV ads here in the UK (none of which have been any good, I hasten to add). Netscape did have 2 TV ads for the launch of Netscape Online. It featured a coupla guys in Baseball caps who were introduced as being from a company that knew the Internet. Quite good from an advocacy point of view - but I guess not as effective as the ad you dismiss. Going right back to my original point, Mozilla/Netscape 6 as a brand name means nothing to joe public. That needs to change. Whether you us a girl in a weird dress or you go for something less "patronising", you still need to build a brand. Otherwise - back to original point - Microsoft Internet Explorer becomes the de facto standard. You say: "Simple, Microsoft will introduce very easy-to-use and slick tags that do very useful and inventive things. Page designers will easily pick them up for sites because 90-95% of surfers use a browser that supports them. Those stuck with Mozilla are either not going to get the full-experience or not going to get any experience." What I'm wondering is.. Why is it that these easy to use, slick tags aren't part of the standards? Does it help the situation that Mozilla's [standards complient] way of doings things is much more difficult than Microsoft's (IE's) way? If IE supports standards AND it's own methods then as long as you, as a user, use IE, all sites will work for you. With Mozilla, they won't. When 98% of all users have an operating system that COULD run IE, and of those some 95% have an OS that came bundled with IE, how many will actually switch? The amount of people who care about IT politics and boycotting Microsoft is very very small. That's the remaining 1% - 2% in the stats.. Because the standards take as a vitally important criterion the separation of content and appearance. If display properties, interactivity behaviors (using the word behavior in its generic sense) and document structure are kept separate, pages will degrade gracefully in other user agents (not just other WinMacNix browsers, but in AvantGo, telephones, etc.). Having these so-called proprietary "tags" (i.e., non-standard elements) breaks the separation of content and appearance and thus can easily lead to missing opr garbled content in some user agents. You are not taking into account the possibility of AOL switching their browser to Netscape/Gecko. If that happens, which is a distinct possibility now that their negotiations with MS broke down in a bitter way, Netscape/Gecko will get a big increase in market share. Another point to consider is that PC web access is becoming saturated and that the next wave of web access is coming from embedded systems, multifunction pagers and cell phones. Mozilla has been chosen already by Nokia and others to embed their browser into their systems. As a mater of fact there are countries in Europe where more people access the web/web emails accounts through their cell phones than from PC. The war is not over but it is moving to another front. Have you actually used IE5 for OSX? At the moment, it's much buggier than Mozilla is. (As I'm sure you know, OSX is also a Unix, but some might think that your message implied that the NeXT-like operating system is a descendant of Mac's System 7/OS8/OS9 family). > and the failure of Linux > companies and organizations > to make a truly consistent > and usable Linux desktop > that works for the end-user, A lot of this is a failure of patience on the part of the VC community. If Eazel had been able to keep the money flowing for a year or two more, we'd see a very easy to use Linux desktop. > So, when we get back to > the main operating systems > that have 97-98% of the > desktop market (WinXX and > MacOS), you see IE being > given not chosen. I have to ask where these figures come from. Sure, among home users and commercial users, these figures could be right (*could be*), but among academic users there's more balance. One certainly can't just count the number of CDs purchased, or the number of desktops which come with a particular system preloaded (despite the methodologies used in some recent, highly-publicized studies). >>Supporting Mozilla and IE5+ with shared code is quite possible and reasonable to do<< Not for anything non-trivial, it isn't, like any kind of DHTML interactivity beyond simple form validation. It's quite UNreasonable to do as things stand now. Duplicating the test matrix is a bad problem all by itself, but in addition, both browsers are extremely quirky. They aren't even the same across their own minor versions or across platforms, much less between each other. We're basically looking at 25% increase in development cost and a 100% increase in QA cost to support both. That just can't be justified by anything except significant market share. Between 85% for one and 0% for the other, guess which one we're going to support? Now, if the Mozilla share got up to, say, 20%, or we had major customers who mandated Mozilla at their sites, that would be a different story. Otherwise, no fricking way. If you write things according to the DOM standards (instead of sticking with IE's or NS4's "DHTML"), most of the time they'll work in both Mozilla and IE5. >>If you write things according to the DOM standards (instead of sticking with IE's or NS4's "DHTML"), most of the time they'll work in both Mozilla and IE5. << If only that were true. In fact, it's only true for extremely simple, toy-level projects and form validators. Try do anything sophisticated like re-sorting tables on the client, designing custom interactive widgets, using layers for drag and drop, and so on, and one quickly finds otherwise. okay, so i've been able to get these compile options out of a google search: ac_add_options --disable-xprint ac_add_options --disable-logging ac_add_options --disable-mailnews ac_add_options --disable-bidi ac_add_options --enable-optimize=-02 ac_add_options --disable-debug ac_add_options --without-qt ac_add_options --disable-tests ac_add_options --enable-strip-libs ac_add_options --enable-x11-shm ac_add_options --disable-static now, can someone tell me what affect --disable-static, --enable-strip-libs, --enable-x11-shm has on the build? also, if i decide not to compile Mail and Composer, does this have any affect on mozilla's performance or just footprint and compile time? feel free to share any other compile options that might be useful. cheers, -Brendon --enable-strip-libs strips the libraries of debugging symbols. Smaller binaries, less useful things you can get out in a debugger. --disable-static is the default right now. It builds with dynamic libraries and runtime linking instead of building one huge binary. "nsHTMLSelectAccessible.cpp in method 'void nsHTMLSelectWindowAessible::GetBounds(nsRect &, nsIFrame **' nsHTMLSelectAccessible.cpp:652 'listControlFrame' is not a member of 'nsLayoutAtoms'." anyone? this is when trying to build the 6th's source with the compile options: --disable-xprint --disable-logging --disable-mailnews --disable-bidi --enable-optimize=-O2 --disable-debug --without-qt --disable-tests --enable-strip-libs --enable-x11-shm --disable-ldap --disable-dtd-debug I build with this and it works great (fear the formatting): BUILD_OFFICIAL=1 ./configure --prefix=/opt/gnome \ --enable-optimize \ --disable-debug \ --with-default-mozilla-five-home=/opt/gnome/lib/mozilla \ --enable-strip-libs \ --disable-tests \ --disable-short-wchar\ --enable-nspr-autoconf \ --with-extensions \ --without-mng \ --enable-crypto This is straight out of the .spec file from Blizzard's RPM. I used to create custom builds without Mail. (I couldn't separate out the editor component back then... I'm not sure if it's possible now.) Someone on IRC told me it didn't make a difference as far as performance. The performance increase I personally observed could have been psychological or minimal due to my optimizing it for i586. So yeah, I think just the compile time is affected. Here's my mozconfig from my compile scripts. # Options for 'configure' (same as command-line options). #ac_add_options --disable-mailnews ac_add_options --disable-debug ac_add_options --enable-strip-libs ac_add_options --enable-cpp-rtti ac_add_options --enable-optimize ac_add_options --enable-mathml ac_add_options --disable-tests ac_add_options --enable-crypto Whenever I start Moz 0.9.2 I get an Error dialog entitled "NSPR:EventReceiver:mozilla.exe - Entry Point Not Found" with the text "The procedure point ?ToNewUTF8String@@YAPADABV?$basic_nsAReadableString@G@@@Z could not be located in the dynamic link library xpcom.dll". I click [OK] and then startup appears to complete successfully and the browser appears to function normally. What's going on? I don't exactly know what's going on, but i think i can tell why it's happening. The same problem occured to my, because i usually download nightly and install them in the same directory than the previous one, without deleting the old one before. if you first delete the old directory and the unpack the new one within the same path it works fine. be careful not to delete your user data, but on a standard win system they are saved in a totally different place anyway (as far as i remember) I clicked on the IE icon, then click on the Mozilla and both browser windows appear at the same time!! However, it's a different story about opening new browsing windows. While IE open instantly like any other window, Mozilla open the new window like when the first windows launch. The differences is not that big but mozilla is noticibly slower. As for those who wonder what my system spec is, there you go: K6-2 350 196 MB ram Running on Win2K and the IE I compare to is IE 5. Well, I'm happy with 0.9.2 because now I can finally watch quicktime movies with the plugin! For some reason, IE 5.5 doesn't use the plugin... weird. Anyway, at least there is one thing that mozilla does that IE doesn't do on my computer! :) Scott I think you are still missing the point. No that long ago, a very large, heavy industry parts manufacturer here in my region began an aggressive television advertising campaign that ran during one and only one particular show. It was one of those ads that left you scratching your head and asking 'who on earth would be interesting in this stuff.' Well, it was later reported to be a concerted effort to reach a single (repeat: a single) decision-maker who, this company knew, was a fan of the show. Likewise, the company for which I work has many individual clients that represent more revenue than the cost of maintaining a cross-browser e-commerce site. If any one of them called to say their choice of browser did not work with our site, the current team of programmers would be executed and a new team of programmers would be hammering out a solution before the client could hang up the phone! Just because 99% is happy does not mean you can ignore the 1%. As I said before, 1% can represent $(m/b)illions!!! When dealing with large enities or billions of smaller ones, 1% represent a healthy sized market from which many wealthy people/companies have made their riches! Under Windows 2000 (SP2)I downloaded and installed mozilla-win32-0.9.2-installer.exe, but evidently there was no MathML support there. For example, the first math display at http://pear.math.pitt.edu/mathzilla/Examples/Fermat/flt01.xml appeared as follows (where I've had to indicate some characters by name, e.g., "para" for paragraph sign, "leftarrow" for a left-pointing arrow, "dbledownarrow" for a downward-pointing double-shafted arrow): The equation para\ leftarrow dbledownarrow leftarrow dagger\ It should start as follows (where I've used the notation x_n to mean x with a subscript n, etc.): The equation x_n + y_n = z_n So I uninstalled that version of Mozilla, downloaded mozilla-win32-0.9.2-MathML-SVG.zip, unzipped, and tried that. Exactly the same problem. Are there fonts missing that I need to install, or what? They worked under .91, but fail under .92 (sometimes the browser crashes, and sometimes there's a message about a file in the cache not being found). In general, I've found .92 to be a bit less stable than .91 I'm a developer who recently made the switch to Mozilla from MSIE (I even use the mail now, having migrated from OE). I'd been downloading milestones and filing bug reports here and there since M14 (not so long ago I suppose) and looking forward to one day being able to make the switch. I've been using 0.9.2 as my default browser for a few weeks now and love it. While I'm looking forward to fixes for the bugginess of the bookmarks system and the history feature, I don't see myself going back to MSIE anytime soon. I just wanted to express my gratitude to the Netscape engineers and the active non-NS contributors for turning out such a kickass piece of software. Thank you. |