| eructate | typically, if the system does claim to support interrupt-process, it includes code in loops to poll for interrupts. because it's usually not safe to smash a new stack frame onto a process. |
| mega1 | no asynchronity => no useful C-c (?) => my lisp image is not the operating system I though it to be => dunno I'm obsessed with C-c maybe someone has a cool idea how to avoid needing it |
| nikodemus | mega1: you can have it ;-) |
| ignas | mega1: C-c rehab? |
| nikodemus | mega1: ...as long as i can make it synchronized ;P ...for my code |
| xinming_ | when I am in clisp, How can I check a documentation for a function please? |
| mega1 | tell me instead what's needed to implement no-shared-heap thread in sbcl |
| rudi | xinming_: (describe 'the-function) might be the easiest way |
| xinming_ | thanks |
| nikodemus | mega1: at any rate, i don't think i'm getting to the indefinitely deferrable interrupts for awhile yet. with-deadline is the important bit mega1: i was hoping you'd tell me... mega1: i suppose you get to implement your own copy-on-write? |
| mega1 | ah, that hasn't occured to me. I was thinking in terms of marshalling messages. and lamenting the loss of the shared global binding |
| nikodemus | otherwise, how do you prevent (setf (car *foo) t)? |
| mega1 | by giving distinct heaps to threads? |
| xinming_ | But the problem now is, It will connected over the inter if we use (describe 'fun-name) Is there a short summary for this? |
| nikodemus | how do you give them distincs heaps? by copy-on-write, of course! :) |
| rudi | nikodemus: you assume that the new thread wants to have shared data with its parent hmm, otherwise there won't be any functions etc in the new image - you might have a point there ... |
| nikodemus | rudi: you could do copy-on-write from the original core too oh! |
| rudi | well, a spawned worker thread might not want to start its life by reloading the server code it's supposed to be executing :) |
| mega1 | how to "share" and propagate changes to mutiple thread is an issue, erlang can have an old version of a module too |
| nikodemus | you could do (publish-image :foo) in a thread that has loaded all sorts of stuff, and then (make-thread 'server-function :image :foo) that might be fun ;-) |
| antifuchs | haha, extremely heavyweight threads (-: |
| knobo | does anyone use sbcl on solaris? aha, now I see. There is no port to sparc |
| nikodemus | there is the binary is a little old, though but there should be no fundamental issues with it |
| antifuchs | solaris 10 on x86 is probably where it's at these days |
| dan_b | if you have a lisp-is-the-os view, perhaps you should map C-c into that view as the SAK key |
| nikodemus | it is just a bit finicky about the toolchains (gnu vs sun) |
| dan_b | "stop what you're doing and drop me in the hypervisor" |