Firefox 3.5: Stumbling out of the gate

Tuesday 30 June was the day that Firefox 3.5 was officially released. Many sites sang glorious hymns to its new features and overdue improvements. Based on those initial reports (I'm so gullible) I went slumming over to and downloaded the 3.5 installer. Know ye that I am a unrepentant Googlite, preferring to worship at the shiny altar of Chrome.

I've used Firefox for a long time, helping to use and test it when it was first known as Phoenix (remember them big ugly orange buttons?). Light and lean when compared to Mozilla, it was just what I wanted and needed for my own personal use.

Time marched on, and Firefox accreted features and bloat. In particular it became a memory hog around version 2. I left Firefox at version 2 on my Windows notebook, while upgrading to version 3 under Linux (both willingly as well as part of the general release schedules). I was never really tempted to move from 2 to 3 under Windows; after all, It Worked For Me and that was all that mattered.

Then, in September 2008 I installed the first Google Chrome beta for Windows, and I never looked back. It combined awesome stability with awesome speed and awesome simplicity. I wound up with not one, but two legacy browsers on my Windows notebook; IE 7 and FF 2.

Smoke Test

Based on the glowing review at Ars Technica, I downloaded and installed 3.5 earlier today. FF 3.5 installed without any issues, upgrading my NoScript plugin during the installation process. That is the first time I've every upgraded Firefox and have NoScript properly tracked, so kudos to both teams on that. The new FF also remembered all my open tabs and other bits. From an installation perspective it was fast and absolutely flawless.

Problems occurred during execution. The Ars Technica article points to a new feature developed for Firefox, 3D transforms. When I executed the demo the entire browser crashed.


One of the three reasons I gave earlier for switching to Chrome was its stability, which is due in no small part to the design decision to use a multi-process architecture, where a process is assigned to each site instance and plugin. Crashes in tabs I can live with. Crashes of the entire browser due to a problem in one tab is no longer excusable, especially within the first 30 minutes of trying out a new release.

I'm in the process of building a proof-of-concept in which certain 'fat' clients are replaced by rich browser applications. One of the assumptions going into this was that I could find at least one rock-solid browser that would not crash if there was an issue in a tab. So far Chrome lives up to that assumption. I had high hopes that FF 3.5 might be as stable, and thus a viable alternative. But after today's minor adventure I'm not so sure. At least I have Chrome.


Popular Posts