Mozilla 1.0 Not Ready for Prime Time -- Close but No Cigar and No Brass Ring!
A Quick Look at Mozilla 1.0 Browser-Suite Performance -- Speed, Stability, and Memory Hogging
Mike Angelo -- 18 June 2002 (c) -- Page 3
The focus in this article is on Mozilla 1.0 performance in the Microsoft Windows environment. Primarily, this is because Microsoft's Internet Explorer browser does not run natively in the Linux environment. Also, Stylesoft's NetCaptor browser caught our attention recently and we wanted to compare Mozilla performance to NetCaptor performance. NetCaptor is a Microsoft-Windows-only browser product.
The emphasis of that focus is on the effects and impact of Mozilla 1.0 performance on the complete user experience. That's why our testing looked beyond just page-rendering speed. Page-rendering is just one part of the overall user-experience with a browser. Opening browser-window sidebars, checking the in-program help, changing view parameters such as text size, bringing a browser up from the task bar to the desktop, and so forth all are important parts of the complete user-experience (CUE).
An extra second or two to open a sidebar, a few seconds more to open a Web page, extra time to start the in-program help, more time to change text-size -- more time than it takes to do all that in another browser -- add all those little time-slices together and multiply them by the number of times you do those things while Web surfing -- seconds turn into cumulative minutes, cumulative minutes turn into cumulative hours. Before you realize it, you have wasted hours and hours of your accumulated time because Mozilla is sluggish on machines that do not count their speeds in the GHzs instead of the MHzs.
Additionally, how much a browser sucks up system resources is important to the user-experience too -- particularly in how sucking up resources impacts on using other applications. Typically, many computer users have several applications open at the same time (multitasking) and switch among the several applications as they use their computer.
If an application sucks up lots of system resources, that can limit or restrict how many other applications you can have open at the same time. Low resources also can affect the performance of several applications running on your computer -- not just the application that is hogging the resources.
Both Internet Explorer and NetCaptor performed noticeably better speed-wise than Mozilla performed in our informal, side-by-side browser tests -- on a 300-MHz machine with 128-MB RAM running Windows 98 SE. However, for the most part speed-performance differences among the three browsers informally tested on a 1-GHz box (with 512-MB RAM running Windows 98 SE) were not remarkably noticeable.
A very noticeable exception to that, however, is the all-too-long-a-time it takes Mozilla 1.0 to launch. The Mozilla developers blame the remarkable difference in how much longer it takes Mozilla to launch compared to the much shorter time it takes Internet Explorer to launch on Internet Explorer being well-integrated into the MS Windows operating system and desktop. While that might have some effect, it is more likely that having to load the Mozilla Application Programming Framework plus the Mozilla browser-suite application in order to get the Mozilla browser up and running is an important, if not a major, factor.
The Mozilla developers added a Quick-Launch feature to Mozilla in an attempt to cover-up that lengthy launch time. However, we do not recommend using the Quick-Launch option. It loads the Mozilla APF whether you need it or not -- thus sucking up resources and adding instability to your system unnecessarily when you do not have the Mozilla browser-suite open.
The Mozilla folks also call Quick-Launch turbo mode. For a more thorough discussion of Quick-Launch, please see our article Turbo Mode & Bugs Slow Mozilla Development to Snail's Pace.
Overall, the Mozilla browser and e-mail modules are remarkably slow and sluggish on our 300-MHz test box running Windows 98 SE.
Mozilla still does not seem to handle Windows memory as well as it should. This can be particularly significant when running Mozilla on Microsoft Windows 9.X which is notorious itself for poor memory and multi-tasking management.
The Mozilla crash problems we experienced in our evaluations appear to arise from Mozilla's memory consumption problems. With plenty of Web pages opened in a combination of Mozilla windows (and tabs opened in those windows) Mozilla easily can bring free system resources down to zero -- thus crashing either Mozilla or the entire system.
Application lock-ups and system crashes steal your time too. They steal the time it takes you to restart the application or reboot your computer. They steal the time it takes to get everything back to where it was before the lock-up or crash -- including the time it takes to recover or re-do documents and other projects that are lost or damaged by application lock-ups and system crashes. Currently, there are nearly 150 open, unfixed dataloss bugs and more than 600 open, unfixed crash bugs listed in the Mozilla bug-tracking database, Bugzilla.
Although Mozilla's speed-performance problems are not remarkable on higher performance Windows-based computers, the underlying causes for Mozilla's noticeably poor performance on slower machines suggests that Mozilla is not the best choice, performance-wise, for faster machines either.
All-in-all, the Mozilla 1.0 browser-suite does not offer any compelling, performance reason for people to switch from Microsoft's Internet Explorer browser to AOL-Netscape's Mozilla browser. On the basis of overall browser look, feel, and performance, Microsoft's Internet Explorer browser still is a better choice than AOL-Netscape's Mozilla 1.0 browser.
Moreover, so far NetCaptor looks OK and it has the Tabbed-Browsing and ad-blocking features found in Mozilla, which are not found in Internet Explorer. In the speed tests discussed here, NetCaptor outperformed Mozilla. We will be taking a closer look at NetCaptor at a later time.
In the Linux arena there is a nice assortment of Web browsers available, such as KDE's Konqueror browser and file manager. Konqueror is an integral part of KDE. At this time, we see no compelling reason to use the Mozilla browser instead of the Konqueror browser or other popular Linux browsers.
Nevertheless, Mozilla 1.0 is a nice and usable product. Moreover, although Mozilla 1.0 is not rock-solid, it is relatively stable and reliable. Additionally, the Mozilla browser-suite does have many nice and handy features too.
Certainly Mozilla 1.0 is the best Mozilla offering to date even though it also is the buggiest Mozilla ever. If you already are a Mozilla user, you ought to try Mozilla 1.0 if you already have not done so.
If you have not tried Mozilla, you ought to give it a test spin. Mozilla 1.0 is easily downloaded from the Mozilla Organizations FTP server(s) and it is easy to install.
However, the bottom line is that Mozilla 1.0 is premature. It still has too many performance issues. There are too many annoyances, bugs and issues that need to be resolved before Mozilla is ready for prime time. Mozilla 1.0 gets no cigar and it gets no brass ring!
Incidentally, please check the MozillaQuest Magazine front-page (mozillaquest.com) sidebar every now and then for Mozilla bug-count updates and for post-Mozilla 1.0 progress updates. MozillaQuest Magazine will have additional Mozilla 1.0 release information as it becomes available. Please check the MozillaQuest Magazine front page regularly for post-Mozilla 1.0 milestone and update release news too.
Essential XUL Programming, by Vaughn Bullard, Kevin T. Smith, and Michael C. Daconta, John Wiley ISBN 0-471-41580-4, $45
Memory leaks bug -- Bug 81446
The stuttering bug -- Bug 91643