Lars T Hansen

@jfbastien, maybe you can write more about what specifically you expect out of "per-thread init" and what the use cases look like, so that we can move this issue along....

The Nov meeting notes record this exchange: BS: Additional proposal that came up last time about initialization - turning that off and on JF: Dynamic linking? BS: Being able to...

@rossberg, generalizing it further, it is basically a memcpy from a static source outside the heap, is it not? If we're going down that road, a flag in the segment...

I'm in the process of implementing this (not *quite* done yet but close enough) and here's the design I've ended up with. The SAB object holds the length of the...

No. For shared memory we reserve the memory up to the max (or to the usual 6GB range on x64) and just commit as the memory grows.

OK. After discussing this with Luke I'm coming around to the idea that even reading the SAB's length field should acquire the lock, this will tighten things up in the...

@jfbastien, I don't think a signal handler would be required in any sense. Everything can be done with explicit bounds checks if necessary and as Ben says, if other threads...

Sorry about that. Sometimes I even confuse myself :) Thinking about this a little more, consider an implementation that has explicit bounds checks and a non-moving buffer where the memory...

What's currently implemented in Firefox (and available in Nightly behind the javascript.options.shared_memory flag in about:config) is roughly what I outlined before, namely, we take a lock for memory.grow and also...

I thought the process required the spec text but it does not, so we don't have to wait on the memory model for phase 3, for one thing. (FF, like...