freenode/#sbcl - IRC Chatlog
Search
20:20:55
phoe
I'm trying to look into https://github.com/cosmos72/stmx/blob/9a704e142216015783e8bb1f8d7cb8d90a14f6f9/asm/x86-32%2C64-insts.lisp#L134 not building on 1.4.10+
20:21:26
phoe
And it seems that this patch is kind-of trivial, since we want to emit a NOP instruction if I read the comments correctly.
20:24:49
AeroNotix
I suppose emit-skip uses the current ISA to lookup waht the NOP opcode is and nothing else/
20:29:48
pfdietz
An agreement on what users need, and what the sbcl devs are willing to provide. Just grabbing internal symbols is a source of endless breakage.
20:30:10
AeroNotix
pfdietz: yeah this was using an internal symbol, no fault of SBCL. Just trying to figure out how to fix it
20:30:41
phoe
pfdietz: I'm looking for a fix more than for an API - there's too many ::s in there for me to start doing anything about it right now
20:31:00
pfdietz
That users ARE grabbing internal symbols indicates it could be useful to provide something for them.
20:31:49
phoe
https://github.com/cosmos72/stmx/blob/9a704e142216015783e8bb1f8d7cb8d90a14f6f9/asm/x86-32%2C64-insts.lisp is full of SB-VM code that does assembly breadboard
20:31:51
Bike
in this case it's doing so because it's hooking into sbcl's code generator, which might be hard to keep up support for
20:35:42
phoe
renaming sb-vm::emit-skip to sb-vm:emit-skip doesn't fix the fact that the symbol EMIT-SKIP is nonetheless not found in the SB-VM package
20:36:24
AeroNotix
phoe: EMIT-SKIP was implemented in terms of a public API though. I've "borrowed" the original EMIT-SKIP code and it gets past that error now