Fcalva/cyclone
1
0
Fork 0
mirror of https://github.com/justinethier/cyclone.git synced 2025-05-24 20:45:06 +02:00
Code Issues Projects Releases Packages Wiki Activity Actions

Added notes

Browse source
This commit is contained in:
Justin Ethier 2021-01-18 16:12:54 -05:00
parent dfdad96d57
commit 5ad0f6fa68
1 changed files with 16 additions and 0 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

16
docs/C-Calling-Conventions.md
View file

@ -96,6 +96,13 @@ Disadvantages:
Based on this overview I am planning to use C arrays rather than Cyclone vectors to pass arguments. Based on this overview I am planning to use C arrays rather than Cyclone vectors to pass arguments.
### Packing Arguments
don't even need alloca for this, can just use an array, EG:
object aa[2];
aa[0] = arg1;
aa[1] = arg2;
TODO: how to call these functions, need to pack args prior to call TODO: how to call these functions, need to pack args prior to call
TODO: how to unpack args for a call. I think it would be simple, need to change compiler to point to members of `args` instead of directly to arguments TODO: how to unpack args for a call. I think it would be simple, need to change compiler to point to members of `args` instead of directly to arguments
@ -105,6 +112,15 @@ Varargs functions still expect to receive extra arguments as part of a list. we
TODO: example of packing/unpacking for fixed args, EG: (lambda (a b c)) TODO: example of packing/unpacking for fixed args, EG: (lambda (a b c))
TODO: example of packing for (lambda X) and (lambda (a b . c) variadic forms TODO: example of packing for (lambda X) and (lambda (a b . c) variadic forms
### Unpacking Arguments
Unpacking is just simple array references:
arg1 = args[0];
arg2 = args[1];
TODO: any compilications here due to assumptions by our compiler, and how it compiles identifiers?
## Performance ## Performance
A key consideration is the impact to runtime performance. We do not want this solution to significantly slow down our programs. I expect there may be some performance hit though. A key consideration is the impact to runtime performance. We do not want this solution to significantly slow down our programs. I expect there may be some performance hit though.