Advance the port all the way to the end of 2022. #1
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This was referenced Jul 13, 2023
Merged
eddyb
force-pushed
the
forward-the-flotation
branch
from
64f8cba to
339c8f3
Compare
eddyb
changed the title
eddyb
force-pushed
the
forward-the-flotation
branch
from
d7c5328 to
6b86e72
Compare
eddyb
changed the title
wesleywiser
approved these changes
Jul 18, 2023
(last APFloat-related LLVM commit from 2017).
(last APFloat-related LLVM commit from 2018). This commit is also notable as the last APFloat-related commit before the LLVM relicensing in early 2019 (llvm/llvm-project@2946cd7).
(last APFloat-related LLVM commit from 2019).
(last APFloat-related LLVM commit from April 2020, before BFloat16 was added).
(last APFloat-related LLVM commit from June 2020, concluding the addition of BFloat16).
…nd `f64`/`Double`).
(last APFloat-related LLVM commit from 2020).
…es SNaN->QNaN itself.
…hide known differences.
(last APFloat-related LLVM commit from 2021).
(last APFloat-related LLVM commit from 2022).
…nd other fields.
eddyb
force-pushed
the
forward-the-flotation
branch
from
6b86e72 to
02bb6cf
Compare
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Note: I've had these in separate branches, then in one branch, but I think it's best to submit these as a PR for ease of linking and discussion.
Licensing note: first two "advance" commits are pre-LLVM-relicensing, the rest are after.
Note: this is a copy of the relevant parts of my update comment (rust-lang/rust#55993 (comment)).
Quick rundown of notable commits:
(thanks to upstream fix llvm/llvm-project@f2c2851)
APFloatchanges before the massive LLVM relicensing (llvm/llvm-project@2946cd7) in early 2019(thanks to upstream fix llvm/llvm-project@e62555c)
(as the port now includes llvm/llvm-project@e34bd1e and llvm/llvm-project@149f5b5 without selective backporting)
rustc_apfloatdoesn't take certain questionable shortcuts, this exposed a testing mistakeall the way back from 2013: APFloatTest: convert test accidentally ended up with
x87DoubleExtended->x87DoubleExtendednoops. llvm/llvm-project#63842(thanks to upstream fix llvm/llvm-project@ed6c309)
bruteforce-tinysubcommand for exhaustively checking 8-bit formats.APFloat's
addOrSubtractSignificandstill throws assert in subtler FMA cases. llvm/llvm-project#63895f32::mul_add/f64::mul_add(including LLVM constant-folding: arustcbuild w/ LLVM assertions enabled will crash)rustc_apfloatwill always trip theassert!but LLVM can just give wrong resultsAt this point I think it's safe to say this version of
rustc_apfloatis strictly superior to what's incompiler/rustc_apfloatright now (okay, sure, maybe it has more typos, and we still need to bring in @wesleywiser's detailed documentation of all the licensing history, before it can actually be used by anyone).