Using Flock was kind of a long shot. Flock is based on Gecko, like Firefox, which has given me a lot of problems. Flock served me well at first, but then starting gobbling up RAM. So, I’m pretty sure I’ve narrowed it down, for me, to XUL and extensions.
I’m not sure exactly what’s to blame for the specific problems I have: other programs on the computer? a single bad extension? I don’t know, but whatever it was, it was present in FF1.5, FF2, and Flock. And each had their own set of extensions installed. It’s not a core Gecko problem, because, as I showed before, Camino doesn’t have the same problem.
So, at the urging of Nate, who, I guess, also spends some time at his computer, I decided to go for a full time ride at Opera.
I don’t require THAT much from a computer: mainly, it has to be able to sustain my browser requirements. And those are tough, because I expect to be able to open 10-15 tabs and still have the browser function without (a) eating up greater than 200MB of RAM, (2) eating up > 10% of the CPU for more than a few seconds, and (third) locking up the browser or worse, the entire system. Enough use of Gmail, Flickr, or other AJAX apps and my Windows Gecko/XUL browsers toast themselves and everything around them. So I’m giving Opera a go.
The only crappy thing is that there’s no way to import form cookies, form history, cookie block list, ad-block filters, or history. And that sucks, because it’s going to take me a long time to rebuild that.
In my experience Camino exhibits memory leaks and if left open for a couple of weeks will eat much of system memory. Once that happens, the browser will hang whenever the automatic garbage collector is invoked. Eventually the browser’s garbage collector will thrash the system since traversing the object graph requires constant memory paging. Since a garbage collector is used, the likely source of many of the memory leaks is the failure to nullify a pointer when the object to which the pointer points is no longer needed. There is something of a misconception that in systems with automatic garbage collection, objects do not need to be released; in reality, the way to release an unneeded object is to nullify the referencing pointer.
Camino exhibits memory leaks and if left open for a couple of weeks
A couple of… WEEKS? That’s fine by me!
FF and Flock exhibits problems after several hours of use. A single, uninterupted Camino session works for me for weeks – regularly. Optimistic or not, that’s pretty damned good!
IMHO, you suffer from low expectations. 🙂
BTW, I have observed that when Camino gets sufficiently bloated to start crushing the rest of the system to death, it seems to have a large number of (possibly zombie) JVM threads. Certainly, the process owns many more threads than it does when it starts up. Also, when I sample the process (i.e. select View > Sample Process from the Activity Viewer application), most of the threads seem to be stalled (i.e. every frame in the thread’s stack has the same frequency count implying that thread has not executed during the sampling period) inside the JVM (or at least, I am assuming that is the case from the symbol names prefixed with “JVM”). It is possible that zombie threads with undead JVMs might be responsible for much of the memory leakage. Just a theory, but I have experienced other problems with the JVM plugin for Camino.
Do you observe similar behaviour with Camino? I’d like to gather enough information to file a bug report, but at the moment I don’t really have anything specific enough to report.
Oops, forgot to put my name on that last post. Duh.
BTW, is Small Axe injecting the “ character into the Subject: line before the `’` character? It’s rather annoying.
Well something is regular expression challenged. The backslash character “ appears okay in the preview, but disappeared from the last post. If it appears in this post, that is because it is escaped by a second backslash. Here goes nothing.
The backslash character appears okay in the preview, but disappeared from the last post. If it appears in this post, that is because it is escaped by a second backslash.
Yes, thanks it’s a bug that I have been lazy about fixing. I’ll take care of it this weekend for sure.