On 25 Apr. 2017 02:51, "Andres Freund" <and...@anarazel.de> wrote:
On 2017-04-24 11:08:48 -0700, Andres Freund wrote: > On 2017-04-24 23:14:40 +0800, Craig Ringer wrote: > > In the long run we'll probably be forced toward threading or far pointers. > > I'll vote for removing the windows port, before going for that. And I'm > not even joking. Just to clarify: I'm talking about far pointers here, not threading. Yeah, I'm pretty unimpressed that the accepted solution seems to be to return to the early '90s. Why don't platforms offer any sensible way to reserve a virtual address range at process creation time? It looks like you need a minimal loader process that rebases its self in memory if it finds its self loaded in the desired area, then maps the required memory range and loads the real process. Hard to imagine that not causing more problems than it solves.