[perl #29261] [PATCH Wrap Out Parameters in PMCs]
Chromatic
# Please include the string: [perl #29261]
# in the subject line of all future correspondence about this issue.
# <URL: http://rt.perl.org:80/rt3/Ticket/Display.html?id=29261 >
Following up on the idea Leo and I discussed this morning, here's a
patch that expects integer out parameters to be wrapped in some sort of
INTVALy PMCs.
This solves my problem rather nicely.
One test in t/pmc/nci.t needed patching and I added a test.
If this is acceptable, we should do the same for float and string out
parameters too.
Please note that this only affects integers passed to NCI subs via
pointers.
-- c
Leopold Toetsch
> Following up on the idea Leo and I discussed this morning, here's a
> patch that expects integer out parameters to be wrapped in some sort of
> INTVALy PMCs.
> This solves my problem rather nicely.
Good. As soon as we have Dan's ok and I'll have JIT adapted too, it'll
go in.
leo
Dan Sugalski
Works for me. I was going to have you guys work out what you need and
put it in, but you've beaten me to it. :)
If it alters the way things look to bytecode, just make sure to
update the docs and patch up the existing wrapping libraries.
--
Dan
--------------------------------------"it's like this"-------------------
Dan Sugalski even samurai
d...@sidhe.org have teddy bears and even
teddy bears get drunk
Leopold Toetsch
> If it alters the way things look to bytecode, just make sure to
> update the docs and patch up the existing wrapping libraries.
Docs are adapted. Needed code changes:
old for signature "i3i":
I5 = arg1
I6 = arg2
new
new P5, .PerlInt
P5 = arg1
I5 = arg2
leo
Leopold Toetsch
> Following up on the idea Leo and I discussed this morning, here's a
> patch that expects integer out parameters to be wrapped in some sort of
> INTVALy PMCs.
Thanks, applied + changed '2' and '4' too + JIT/i386.
leo
Chromatic
> If it alters the way things look to bytecode, just make sure to
> update the docs and patch up the existing wrapping libraries.
I'll take a look at Leo's new docs and patch everything with an
offending signature in src/call_list.txt.
Anyone who has NCI libraries that pass pointers to primitives, drop me a
note and I'll help you fix them.
-- c