cfbolz changed the topic of #pypy to: PyPy, the flexible snake (IRC logs: https://quodlibet.duckdns.org/irc/pypy/latest.log.html#irc-end ) | use cffi for calling C | if a pep adds a mere 25-30 [C-API] functions or so, it's a drop in the ocean (cough) - Armin
<muke>
cfbolz Hey, I'm a bit stuck getting the finish op to work, the return value doesn't seem to be getting stored properly so when the test reads it it just gets back the first argument instead
<muke>
I've looked at the IR I'm generating and it all seems fine, do you have any idea of what within pypy could be happening? I've been writing into the jitframe pointer passed to the function and then just returning that pointer again, do I need to maybe allocate a new struct or something?
<cfbolz>
muke: I'm not really around right now, but can you paste the ir? Also, does passing arguments *into* the trace work correctly?
<muke>
cfbolz oh no worries, if you like I can detail things in an email and you can share your thoughts when you have the time, there are other things I can work on the meantime too so no rush
<cfbolz>
muke: no, please show me, I'll be afk the whole next week
<muke>
here's the IR
<muke>
ah alright
<muke>
yea I can't think of any reason the argument passing would be messing up either but I'll try and verify it from within llvm too
<cfbolz>
muke: yes, I'd try that
<muke>
cfbolz looks like it is to do with what's being passed, if I try and return the input arg read from the jitframe in llvm, pypy segfalts
<muke>
weirdly I can still compare against whatever value it is within llvm, but it still isn't 2 like it should be
<muke>
that being said, the descr field in the jitframe can still be accessed and holds the correct result, so it's the arg array specifically somehow instead of the whole jitframe