689 commits

Author	SHA1	Message	Date
Ayaz	cef2c8ae87	Update build.rs ... Signed-off-by: Ayaz <20735482+ayazhafiz@users.noreply.github.com>	2024-09-21 09:50:17 -04:00
Ayaz	9c656dd28e	Merge branch 'main' into ayaz/optimize-list-literal-alloc	2024-09-21 09:44:50 -04:00
Luke Boswell	49dedf0b7b	Merge branch 'main' into ayaz/optimize-list-literal-alloc	2024-09-17 19:44:02 +10:00
Brendan Hansknecht	4e3ec97d7a	generate array of alignment type to enforce tag data alignment	2024-09-16 17:10:49 -07:00
Brendan Hansknecht	a7eae239f0	Avoid explicit tag padding ... Explicit padding changes the c-abi when passed to/returned from functions. This is leading to incorrect loading of types like `Task U64 {}`.	2024-09-16 16:47:20 -07:00
Brendan Hansknecht	7997cf95e7	ensure alignment is never 0. Must be at least 1.	2024-08-25 11:48:46 -07:00
Brendan Hansknecht	cf5cd84d60	fix alignment of 16 bytes	2024-08-21 09:04:07 -07:00
Brendan Hansknecht	6b04318e27	handle float types	2024-08-09 19:27:47 -07:00
Brendan Hansknecht	462872551e	fix up list literal constant loading	2024-08-09 19:07:01 -07:00
Ayaz Hafiz	9abbcfaafc	Optimize literal list construction in LLVM backend ... Currently, list literals are always heap-allocated and their elements are stored by emitting a GEP and store for each item in the literal. This produces huge quantities of IR, causing compile times for e.g. programs with large literals or ingested files to blow up. Instead, if a list literal consists entirely of literal values, create a global section for the literal and return a pointer to it.	2024-08-09 18:51:17 -07:00
Brendan Hansknecht	82764dc697	switch to passing the copy function into zig (implement llvm)	2024-07-21 19:04:57 -07:00
Brendan Hansknecht	bccab34833	correct comment	2024-07-20 21:30:25 -07:00
Brendan Hansknecht	e804501a92	clippy	2024-07-20 21:21:43 -07:00
Brendan Hansknecht	703224b275	fix alloca args overwriting eachother	2024-07-20 21:14:31 -07:00
Brendan Hansknecht	c5a74bdc12	for referenced values, use allocas instead of phi nodes	2024-07-20 19:52:11 -07:00
Brendan Hansknecht	ee7f1e39c7	move all allocas to the entry block	2024-07-20 19:17:11 -07:00
Brendan Hansknecht	c734a27b59	move List.map* into roc	2024-07-13 10:39:56 -07:00
Brendan Hansknecht	94cb4fb8ff	remove inc functions from list.map*, the closure will inc in necessary	2024-07-13 10:39:56 -07:00
Brendan Hansknecht	e92cf1be42	Revert "attempt to correct list refcount generation in dev backends" ... This reverts commit 0854a1a53a2db441c648140bdd24de91125ec645.	2024-07-13 10:39:54 -07:00
Brendan Hansknecht	36bc8bfac4	attempt to correct list refcount generation in dev backends	2024-07-13 10:39:54 -07:00
Brendan Hansknecht	48eb9c31a9	update list generated refcounting functions for dev and wasm	2024-07-13 10:39:53 -07:00
Brendan Hansknecht	0d0a4a8806	Update refcounting of list builtins to the new system.	2024-07-13 10:39:53 -07:00
Brendan Hansknecht	f87c32ca7d	clippy	2024-07-13 10:39:52 -07:00
Brendan Hansknecht	9b6798ead1	Correct refcounting check to be recursive through aggregate types	2024-07-13 10:39:52 -07:00
Brendan Hansknecht	951cb1ef3d	Correct list replace bitcode arg order	2024-07-13 10:39:52 -07:00
Brendan Hansknecht	3c842196fa	Update LLVM to properly increment and decrement lists	2024-07-13 10:39:51 -07:00
shua	a2c800452e	add branch on ptr width back ... apple arm tests are failing, maybe this will fix it?	2024-06-10 23:16:33 +02:00
shua	f7bec802c0	PR: rename to List.concatUtf8	2024-06-08 13:36:20 +02:00
shua	33e8a7a439	implement Str.concatUtf8	2024-06-07 19:05:52 +02:00
Brendan Hansknecht	dc539f759d	Improve OptSize implementation in Roc ... 1. Use the llvm `Oz` pipeline. 2. Avoid adding extra optimization passes. This shrunk rocci-bird by ~16%.	2024-05-01 08:25:34 -07:00
Anton-4	3d2a5897e5	roc_bitcode false positive	2024-04-09 20:01:33 +02:00
Anton-4	f085b96f78	removed unused deps, added CI test	2024-04-09 19:47:31 +02:00
Fabian Schmalzried	4066d56b24	fix clippy errors	2024-04-02 00:04:59 +02:00
Fabian Schmalzried	313a3d37e6	move finctions	2024-04-01 21:08:17 +02:00
Fabian Schmalzried	376202c6e8	fix from parts	2024-04-01 21:08:16 +02:00
Fabian Schmalzried	9a91597b27	Windows to parts	2024-04-01 21:08:16 +02:00
Fabian Schmalzried	a1bd36812d	fix for llvm-wasm	2024-04-01 21:08:15 +02:00
Anton-4	ae2506469c	added comment ... Signed-off-by: Anton-4 <17049058+Anton-4@users.noreply.github.com>	2024-04-01 21:08:15 +02:00
Fabian Schmalzried	609a8091c4	remve unused import	2024-04-01 21:08:14 +02:00
Fabian Schmalzried	834015a758	From parts works with llvm	2024-04-01 21:08:13 +02:00
Fabian Schmalzried	e697064ae3	XtoParts works in llvm ... TODO: celanup	2024-04-01 21:08:13 +02:00
Anton-4	35a26eb39f	fix typos, WIP on NumF32ToParts	2024-04-01 21:08:12 +02:00
Fabian Schmalzried	4269608ebf	remove i128OfDec	2024-04-01 21:08:12 +02:00
Fabian Schmalzried	11998b9cc8	f32FromParts and f64FromParts for repl	2024-04-01 21:08:10 +02:00
Fabian Schmalzried	473b8ef2d3	f32/f64toParts in repl only	2024-04-01 21:08:09 +02:00
Fabian Schmalzried	64e9899cad	Add Num.withDecimalPoint	2024-04-01 21:08:09 +02:00
Fabian Schmalzried	9c664172dd	Add Num.withoutDecimalPoint	2024-04-01 21:08:09 +02:00
Brendan Hansknecht	6dc5bfb1b7	Use roc_target over target_lexicon ... Tailors a target class for our needs. Replaces tons of uses across the entire compiler. This is a base for later adding new targets like thumb.	2024-03-31 10:50:26 -07:00
Anton-4	76c47c5d6f	stylistic improvements	2024-03-09 20:47:19 +01:00
Jonathan Schear	504e00d44d	Group toplevel expects by module, run them separately	2024-03-08 14:56:32 -05:00