<wingo>oh that's a bummer (uri abi breakage...) <wingo>i guess if you reference uri? as a value in 2.2 that will break in 2.3 <civodul>the problem is when you compile with 2.2.3 and run with 2.2.2 <wingo>ah that's not supported ever <civodul>i think we've never really thought about it :-) <civodul>though bytecode should remain compatible both ways, right? <wingo>2.2.3 can add instructions that 2.2.2 doesn't have <wingo>in that case it should increment the micro bytecode version and we didn't do that, but oh well <wingo>2.2.2-compiled bytecode should still run on 2.2.3 though <civodul>back to (web uri), i wonder what happens with a 2.2.2-compiled user on 2.2.3 <civodul>essentially it would use the old 'uri?' definition, inline <civodul>which i think "works", except that it doesn't get the uri-reference/uri distinction <wingo>the one thing that might not work would be if a 2.2.2 user used uri? as a value (instead of being applied directly) <wingo>there i think we might have a breakage, as uri? would reference the residualized procedure that define-inline made <wingo>maybe we can make a quick fix though by defining "% uri?-procedure" in uri.scm <wingo>we can release a 2.2.4 soonish <civodul>artyom-poptsov1: i think it would be nice if the remote guile in guile-ssh would listen on a Unix-domain socket under $HOME ***21WAAGR4F is now known as hydraz
<christopher74837>hi, i was wondering: coroutines are super cool, but it seems that there is a huge performance problem with those in guile. I was wondering if that is something that is inherent, or if there is a hope for an improvement some day