Hacker News Reader

Comment by icantthinkofone | Hacker News Reader

In fact, the infamous memory leak Firefox was 'famous' for was fixed in version 2.5 which tells me that anyone who still complains of memory leaks in Firefox is one who you do not want to listen to for advice.

EDIT: I may have the version number wrong, it's been a long time, and, yes, the "leak" was a combination of things and not just one thing but my point is the same.

View on HN

No memory leak was fixed in 2.5, because there was no 2.5 release[1]. What firefox then became famous for was a team of developers who adamantly insisted there were no memory leaks in firefox... right up until some of them admitted there was a problem, and the healing process began[2]. Any software product of this size is going to have SOME memory management problems, and that's fine. Firefox's history of even admitting these exist is spotty at best.

[1] Go ahead and check. I'll wait: https://ftp.mozilla.org/pub/mozilla.org/firefox/releases/ [2] https://wiki.mozilla.org/Performance/MemShrink

What makes you so sure that there was ever only one memory leak? What makes you so sure that one or more new ones haven't been introduced since then?

I've seen and heard a lot of reports from many different Firefox users about Firefox using an unreasonably large amount of memory, even when using fresh installations of the most recent version, and when engaging in very reasonable browsing patterns.

As a software developer faced with a large and frequent volume of reports of such a nature, the only responsible thing to do is to assume that there is truly a problem. This should be assumed even if the developers themselves may be having trouble reproducing the problem. Denying that the problem exists is usually the most counterproductive thing that can be done, because the problem likely does actually exist, and it doesn't get fixed.

By the way, I don't believe that there ever was a Firefox 2.5 release. Perhaps you mean Firefox 3.5?

Users have been reporting memory leaks in Firefox up to and including last week. Users should not be used a source of what a memory leak is cause 80% of users do not know what a memory leak truly is much less be technically competent to recognize and report one.

I often make the error of saying the fix was in version 2.5 but perhaps it was in v1.5 or v2.0. It doesn't matter. They were fixed long ago is my point.

Once again, nobody is denying anything about memory problems. But there are two things at play here:

1. One person's experience is not universal. With a web browser as customizable as Firefox, there will be certain configurations causing problems that aren't caught by tests or by dogfooding. Some people are probably seeing leaks that users with a default configuration don't see. For example: https://blog.mozilla.org/nnethercote/2014/08/15/the-story-of...

2. People need to report their problems in ways that are actionable. HN is not a bug tracker. If somebody on HN (Hacker News! News for programmers who presumably know how a bug tracker works!) is having memory issues, (s)he should have no problems filing a bug in Bugzilla. Firefox has had about:memory for years now; save that report and attach it to your bug.

There have always been memory leaks in Firefox, and before 4.0, they were always worse on a new release, and improved with bug fixing releases. I don't know whether it's aurora, or the increased pace of the development cycle, but I haven't noticed any serious memory leaks like there were back in the day.

Of course, any program the size of Firefox is bound to have memory leaks. So we can't totally eliminate them while we still use C++.

The leak points to a slack development/release process (is there one?). Its expected to check for leaks, crashes before releasing a point build. That this dog got out points to a systemic problem. Was that fixed by 2.5? Or was just the leak fixed.

Reply on news.ycombinator.com