Mozilla Branches for 0.9.8

Wednesday January 23rd, 2002 today branched for Mozilla Milestone 0.9.8, as well as opened the tree for 0.9.9 checkins. Look for branch builds to start sometime later this week, as early as tomorrow. (While a certain site believes that without builds, you can't have a branch, that is not true.) Pre-0.9.9 trunk builds will start tomorrow, or possibly tonight. See our earlier story for what's new in 0.9.8.

#3 A harsh test on 2002012203 build

by dipa

Thursday January 24th, 2002 1:28 AM

You are replying to this message

This build comes a little before branching, so it might characterize the upcoming milestone release too.

More to the point: From time to time, I perform the following test, in order to check browser's stability: I start with site that contains a lot of spam. I open links on new windows. In, this way, I open at least 20-30 browser windows, containing different pages. When the memory usage increases to the point of heavy disk thrashing, I close some windows. Tenths of different sites are browsed.

This test always made NS3.x and 4.x to crash after a few minutes. IE5.5 (security set to medium) performed very well, but after 1-2 hours of testing the system often becomes unstable (not able to save files, open new windows etc). Older Mozilla versions (e.g. 0.9.4) were definitely better than NS4 but not as stable as IE. Build 2002012203 did it (the least to say) astonishingly well. I was able to use it for many hours without a single crash. Memory usage was acceptable. And no system degradation problems, unlike IE. In fact, I never had a so extended browsing session before, on any Win32 os and browser combination. I 'm not sure if this stability was caused by the javascript settings in Mozilla preferences (I disabled pop-up windows, window moves and resizes, status bar overwrites etc). Actually, I don't care, because all these annoying javascript techniques are useless, so any ability to block them is a plus for me.

The above "test case" is not mentioned as a means of comparison between IE and Mozilla. Nor it is a consistent and reproducible test for anyone else except myself (I suppose smoketests are intended for this purpose). Just figured out that current Mozilla is so stable that I could hardly imagine before.

A few big annoyances remain, though: a) Window closing time is still long, when many windows are opened (at least 1 sec). b) Cpu usage is relatively high, enough for delaying new window creation, when the parent window is still loading.

Machine used: K6-III/400, 192MB ram, Win2000.