June 2023 Monthly report

Overview

Thanks again to Open Source Security, inc and Embecosm for their ongoing support for this project.

Milestone Progress

The number of contributions increased this month with over 50 pull requests merged in gccrs from eight contributors and a couple more in our various tooling projects, as well as valuable input on our different platforms, which remains a tremendous help in the development of this compiler. Our two Google Summer of Code students, Mahad Muhammad and Raiki Tamura, are progressing quickly on their projects and submitting high quality contributions - we are getting more and more of Rust’s error codes integrated to our compiler, which will assist us in trying to pass the rustc testsuite, and our lexer is slowly but surely becoming Unicode compliant with a high number of unit tests to boot. Their mid-term evaluations will start next week, and we are excited to enter the second phase of their projects, where more technical contributions will come in: v0 name mangling and punycode support for Raiki Tamura, and internal GCC cleanups for the error emitting code for Mahad Muhammad.

One unforeseen hurdle this month was the need for a complete rework of our name resolution pass. Name resolution in Rust is a complex process, which should be divided in two passes. In gccrs, name resolution was developed before macro expansion, meaning that macro name resolution was not taken into account properly. This is now starting to cause issues as we try and fix the last few remaining bugs around imports, macro exports and early name resolution in general for the compilation of core 1.49. We have spent a lot of time trying to improve the existing name resolution algorithms to stry and split them in two (or more) distinct passes, but this is proving to be a difficult task. Furthermore, it is an area of the compiler which deserves a little more attention, and has not been improved much since its creation. Philip and Arthur took the decision to rewrite this pass of the compiler, aiming for simpler data structures, simpler algorithms and less global state, making the name resolution pass a three-pass process: TopLevel name resolution, Early name resolution and Late name resolution. The TopLevel pass is in charge of collecting definitions for the entire program, as Rust allows you to call a function defined later in the source. Early takes care of macro name resolution and imports, and works in a fixed point fashion with TopLevel. Finally, Late is your “classical” name resolution pass, and resolves functions, types and variables.

We are mostly done with the data structures and are in the process of merging them to our compiler. We now need to write the missing visitors and name resolution 2.0 will be complete.

In our typesystem, Philip spent a lot of time chasing some of the few remaining bugs needed for handling iterators. Most of the fixes are now one-line fixes, which is quite disheartening, but also an indication that we are getting closer and closer!

We are also in the process of welcoming Jakub Dupak’s Master thesis project, which will be focused on adding support for the Polonius borrow checker to gccrs. This exciting work should start in the upcoming months.

Community call

We will have our next monthly community call on the 10th of July 2023. You can subscribe to our calendar to see when the next one will be held. The call is open to everyone, even if you would just like to sit-in and listen. You can also subscribe to our mailing-list or join our Zulip chat to be notified of upcoming events.

• jitsi link
• Calendar ID: 7060a0923ffebd3cb52b1afef35a28ff7b64f05962c9af84c23b1847f1f5f894@group.calendar.google.com
• Google calendar link
• iCal link

Completed Activities

• gccrs: Track fn_once output lang item properly PR2359
• gccrs: fix ICE with inserting autoderef mappings PR2355
• abi: Add testcase for using “Rust” as ABI string PR2353
• Fix type hint propagation PR2352
• Refactoring in prevision of the HIR dump changes PR2350
• Fix bad generic inference on unsafe pointer casts PR2348
• Remove unnecessary methods/fields of `Rust::Lexer` PR2347
• Refactored Field Error Function PR2341
• Fix lexer to skip utf-8 whitespaces PR2339
• Missing tests for utf-8 identifiers PR2338
• [E0027] struct pattern fails to specify struct’s fields PR2335
• ast: Change Identifier definition PR2332
• Match tokens in macros more closely PR2329
• [E0026] Non-Existent Field Extraction in Struct Pattern PR2326
• [E0023] Incorrect Number of Fields in Pattern Extraction PR2325
• gccrs: Stop autoderef of raw pointer types PR2322
• gccrs: Add method selection to operator overloading PR2321
• Fix lexing byte literal PR2320
• Expand expressions/types correctly in more contexts. PR2319
• rust: Add -frust-name-resolution-2.0 option PR2316
• ErrorCode[E0277] Type Does Not Implement Expected Trait PR2315
• Remove `AST::Dump` and replace it with `AST::TokenCollector` PR2313
• TryFrom trait bound issue PR2310
• Refactor lexer to treat all input characters as UTF-8 PR2307
• Prevent invalid iterator dereference PR2305
• ErrorCode[E0433]: Use of Undeclared Crate, Module, or Type PR2301
• gccrs: we can’t check the bounds involving empty placeholder types PR2299
• fixup: Remove old typecheck dump PR2298
• gccrs: fortify resolve_{methodaddress} to match the types PR2297
• derive: Factor common fields inside the base visitor PR2293
• fixup: Remove obsolete error diagnostic PR2291
• Remove direct error on macro rule count-related match failure PR2290
• gccrs: ensure assoicated types are setup for concrete trait impl blocks PR2289
• fixup: Fix build dependency PR2288
• gccrs: introduce new types_compatable PR2285
• Tokenize Unicode identifiers PR2284
• Fix ICE for reference patterns in match statements PR2283
• Fix parsing of statement macros PR2282
• Fix empty nested repetitions PR2279
• Proc macro statements PR2274
• gccrs: Respect the concrete type when resolving qualifed path types PR2271
• Allow parsing a borrow from struct expression PR2270
• Fix warning generated by Optional PR2269
• Handle crate metavariable PR2267
• Fix macro nested repetitions PR2266
• ast: dump literals correctly PR2265
• Add handling of unimplemented derive macros. PR2262
• Make MatchedFragmentContainer a tree-like structure PR2260
• builtin-derive-macros: Add dummy builtin transcriber for builtin derive macros. PR2258
• Handle file loading from “non-mod-rs” files properly PR2227
• fixup: Fix bootstrap build PR2183

Contributors this month

• Charalampos Mitrodimas (new contributor!)
• Pierre-Emmanuel Patry
• Marc Poulhiès
• Raiki Tamura
• Mahad Muhammad
• Owen Avery
• Matthew Jasper
• tieway59
• Dave Evans
• Guillaume Gomez

Overall Task Status

Test Cases

Bugs

Milestones Progress

Note that the intrinsics milestone percentage on github is not representative: It shows a 73% completion rate, but does not take into account the tracking issues with dozens of unresolved items. Thus the percentage is computed using the sum of issues and tracked items done divided by the sums of issues and tracked items overall. Similarly, the Update GCC’s master branch milestone contains a tracking issue containing over 200 tasks. The percentage shown here takes this into account.

Testing project

The testing project is on hold as we try and figure out some of the issues we’re running into with GitHub and our various automations around it.

Planned Activities

• Name resolution v2.0
• Fix remaining type systems bugs for core 1.49
• Look at procedural macro name resolution

Detailed changelog