Scrabsha/html errors #1

scrabsha · 2021-09-07T07:44:08Z

No description provided.

Our goal is to generate HTML pages representing the error emitted by the compiler. This commit adds a -Z html-output flag to rustc. When passed, the compiler will emit its errors in HTML format in stderr. This flag conflicts with any value of the error-format argument. As a bonus, the compiletest crate has been updated to handle the // html-output compiler directive. Adding this flag will invoke the compiler with the -Z html-output flag and compare its output with a corresponding .html file, similarly as .stderr files.

myrrlyn · 2021-09-15T17:26:37Z

compiler/rustc_errors/src/emitter.rs

+                    Ok(()) => {}
+
+                    Err(e) if e.kind() == ErrorKind::WriteZero => return Ok(idx),
+                    Err(e) => return Err(e),


In practice I don't know if it actually matters (I am assuming that you've run tests and observed it work; I haven't tried to work on the compiler before and so haven't actually downloaded or run this myself), but I think this clause loses the current .next() segment if it is not able to write into self.inner, so a retry would skip it. If it works as-is, then I wouldn't worry about it.

Actually it is not a problem if we don't write the current .next(), as long as we exit early and return the correct amount of bytes we actually wrote.

But the good news is that you're highlighting the fact that we should not return an error if we wrote something to the sink. I do need to fix that.

myrrlyn · 2021-09-15T17:36:31Z

Okay so I've played with writing this as a standalone module and while I have some minor practical differences as a result of our prior conversation and seeing your solution first, ultimately I don't think I have anything worth bringing up as an actual change. The Peekable wrapper could be used instead of a local Option, or you could hold the original &str or &[u8] and pull out the SplitInclusive behavior we actually want (.position() and .filter().map()), but these don't affect the API.

I refuse to bikeshed this to a standstill before it even hits rustc. It's behavior I want the compiler to do and you made the compiler do it. Excellent. I think there could be some internal simplification (for instance, I got by without an EscapedHtml enum, and just used (src_bytes_consumed, segment): (usize, &[u8])) but, again, that's getting deep into the weeds when the important part is that it correctly emits text that is safe to place directly into an HTML template.

scrabsha · 2021-09-16T17:16:12Z

@myrrlyn

Thank you so much for taking time to review this piece of code. It's much appreciated.

I tried to remove the remove the trimmed_escaped field and make iter a Peekable<SplitInclusive<...>>, but failed. I think my level in Rust is not good enough to write it cleanly.

I'm wondering how you removed the EscapedHtml enum: how do you know if you must call write or write_all in the Write implementation? Could you post a playground link or something? I don't think I'll implement it though, as the goal is also to stay simple and easy to review.

fix: check against 'main' function name instead of entrypoint function

…-lines-debuginfo, r=wesleywiser tests: Test line debuginfo for linebreaked function parameters Closes rust-lang#45010 which just [E-needs-test](rust-lang#45010 (comment)). To verify that this is actually a regression test, do this, which is a simplified and adapted version of what compiletest does for 1.39 and then 1.88: ```sh for toolchain in 1.39 1.88; do echo -e "\nWith $toolchain:" rustc +$toolchain "tests/codegen/fn-parameters-on-different-lines-debuginfo.rs" "--emit" "llvm-ir" "-o" "/tmp/fn-parameters-on-different-lines-debuginfo.ll" "-g" "-Copt-level=0" "build/x86_64-unknown-linux-gnu/ci-llvm/bin/FileCheck" "--input-file" "/tmp/fn-parameters-on-different-lines-debuginfo.ll" "tests/codegen/fn-parameters-on-different-lines-debuginfo.rs" "--check-prefix=CHECK" "--dump-input-context" "100" && echo OK || echo FAIL done ``` which gives ``` With 1.39: FAIL With 1.88: OK ``` <details> <summary>Click to expand full output</summary> ``` $ for toolchain in 1.39 1.88; do echo -e "\nWith $toolchain:" rustc +$toolchain "tests/codegen/fn-parameters-on-different-lines-debuginfo.rs" "--emit" "llvm-ir" "-o" "/tmp/fn-parameters-on-different-lines-debuginfo.ll" "-g" "-Copt-level=0" "build/x86_64-unknown-linux-gnu/ci-llvm/bin/FileCheck" "--input-file" "/tmp/fn-parameters-on-different-lines-debuginfo.ll" "tests/codegen/fn-parameters-on-different-lines-debuginfo.rs" "--check-prefix=CHECK" "--dump-input-context" "100" && echo OK || echo FAIL done With 1.39: tests/codegen/fn-parameters-on-different-lines-debuginfo.rs:16:16: error: CHECK-SAME: expected string not found in input // CHECK-SAME: line: 10 ^ /tmp/fn-parameters-on-different-lines-debuginfo.ll:69:42: note: scanning from here !10 = !DILocalVariable(name: "x", arg: 1, scope: !5, file: !3, line: 1, type: !9) ^ /tmp/fn-parameters-on-different-lines-debuginfo.ll:69:64: note: possible intended match here !10 = !DILocalVariable(name: "x", arg: 1, scope: !5, file: !3, line: 1, type: !9) ^ Input file: /tmp/fn-parameters-on-different-lines-debuginfo.ll Check file: tests/codegen/fn-parameters-on-different-lines-debuginfo.rs -dump-input=help explains the following input dump. Input was: <<<<<< 1: ; ModuleID = 'fn_parameters_on_different_lines_debuginfo.3a1fbbbh-cgu.0' 2: source_filename = "fn_parameters_on_different_lines_debuginfo.3a1fbbbh-cgu.0" 3: target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128" 4: target triple = "x86_64-unknown-linux-gnu" 5: 6: ``@str.0`` = internal constant [59 x i8] c"tests/codegen/fn-parameters-on-different-lines-debuginfo.rs" 7: ``@str.1`` = internal constant [28 x i8] c"attempt to add with overflow" 8: ``@panic_loc.2`` = private unnamed_addr constant { { [0 x i8]*, i64 }, { [0 x i8]*, i64 }, i32, i32 } { { [0 x i8]*, i64 } { [0 x i8]* bitcast ([28 x i8]* ``@str.1`` to [0 x i8]*), i64 28 }, { [0 x i8]*, i64 } { [0 x i8]* bitcast ([59 x i8]* ``@str.0`` to [0 x i8]*), i64 59 }, i32 13, i32 3 }, align 8 9: ``@__rustc_debug_gdb_scripts_section__`` = linkonce_odr unnamed_addr constant [34 x i8] c"\01gdb_load_rust_pretty_printers.py\00", section ".debug_gdb_scripts", align 1 10: 11: ; fn_parameters_on_different_lines_debuginfo::foo 12: ; Function Attrs: nonlazybind uwtable 13: define i32 ``@_ZN42fn_parameters_on_different_lines_debuginfo3foo17ha98e7c29f4ed8d60E(i32,`` i32) unnamed_addr #0 !dbg !5 { 14: start: 15: %y = alloca i32, align 4 16: %x = alloca i32, align 4 17: store i32 %0, i32* %x, align 4 18: call void ``@llvm.dbg.declare(metadata`` i32* %x, metadata !10, metadata !DIExpression()), !dbg !11 19: store i32 %1, i32* %y, align 4 20: call void ``@llvm.dbg.declare(metadata`` i32* %y, metadata !12, metadata !DIExpression()), !dbg !11 21: %2 = load i32, i32* %x, align 4, !dbg !13 22: %3 = load i32, i32* %y, align 4, !dbg !14 23: %4 = call { i32, i1 } ``@llvm.sadd.with.overflow.i32(i32`` %2, i32 %3), !dbg !13 24: %5 = extractvalue { i32, i1 } %4, 0, !dbg !13 25: %6 = extractvalue { i32, i1 } %4, 1, !dbg !13 26: %7 = call i1 ``@llvm.expect.i1(i1`` %6, i1 false), !dbg !13 27: br i1 %7, label %panic, label %bb1, !dbg !13 28: 29: bb1: ; preds = %start 30: ret i32 %5, !dbg !15 31: 32: panic: ; preds = %start 33: ; call core::panicking::panic 34: call void ``@_ZN4core9panicking5panic17h2f49f09cf859b728E({`` [0 x i64], { [0 x i8]*, i64 }, [0 x i64], { [0 x i8]*, i64 }, [0 x i32], i32, [0 x i32], i32, [0 x i32] }* noalias readonly align 8 dereferenceable(40) bitcast ({ { [0 x i8]*, i64 }, { [0 x i8]*, i64 }, i32, i32 }* ``@panic_loc.2`` to { [0 x i64], { [0 x i8]*, i64 }, [0 x i64], { [0 x i8]*, i64 }, [0 x i32], i32, [0 x i32], i32, [0 x i32] }*)), !dbg !13 35: unreachable, !dbg !13 36: } 37: 38: ; Function Attrs: nounwind readnone speculatable 39: declare void ``@llvm.dbg.declare(metadata,`` metadata, metadata) #1 40: 41: ; Function Attrs: nounwind readnone speculatable 42: declare { i32, i1 } ``@llvm.sadd.with.overflow.i32(i32,`` i32) #1 43: 44: ; Function Attrs: nounwind readnone 45: declare i1 ``@llvm.expect.i1(i1,`` i1) rust-lang#2 46: 47: ; core::panicking::panic 48: ; Function Attrs: cold noinline noreturn nonlazybind uwtable 49: declare void ``@_ZN4core9panicking5panic17h2f49f09cf859b728E({`` [0 x i64], { [0 x i8]*, i64 }, [0 x i64], { [0 x i8]*, i64 }, [0 x i32], i32, [0 x i32], i32, [0 x i32] }* noalias readonly align 8 dereferenceable(40)) unnamed_addr rust-lang#3 50: 51: attributes #0 = { nonlazybind uwtable "probe-stack"="__rust_probestack" "target-cpu"="x86-64" } 52: attributes #1 = { nounwind readnone speculatable } 53: attributes rust-lang#2 = { nounwind readnone } 54: attributes rust-lang#3 = { cold noinline noreturn nonlazybind uwtable "probe-stack"="__rust_probestack" "target-cpu"="x86-64" } 55: 56: !llvm.module.flags = !{!0, !1} 57: !llvm.dbg.cu = !{!2} 58: 59: !0 = !{i32 2, !"RtLibUseGOT", i32 1} 60: !1 = !{i32 2, !"Debug Info Version", i32 3} 61: !2 = distinct !DICompileUnit(language: DW_LANG_Rust, file: !3, producer: "clang LLVM (rustc version 1.39.0 (4560ea7 2019-11-04))", isOptimized: false, runtimeVersion: 0, emissionKind: FullDebug, enums: !4) 62: !3 = !DIFile(filename: "tests/codegen/fn-parameters-on-different-lines-debuginfo.rs", directory: "/home/martin/src/rust") 63: !4 = !{} 64: !5 = distinct !DISubprogram(name: "foo", linkageName: "_ZN42fn_parameters_on_different_lines_debuginfo3foo17ha98e7c29f4ed8d60E", scope: !6, file: !3, line: 9, type: !7, scopeLine: 9, flags: DIFlagPrototyped, spFlags: DISPFlagDefinition, unit: !2, templateParams: !4, retainedNodes: !4) 65: !6 = !DINamespace(name: "fn_parameters_on_different_lines_debuginfo", scope: null) 66: !7 = !DISubroutineType(types: !8) 67: !8 = !{!9, !9, !9} 68: !9 = !DIBasicType(name: "i32", size: 32, encoding: DW_ATE_signed) 69: !10 = !DILocalVariable(name: "x", arg: 1, scope: !5, file: !3, line: 1, type: !9) same:16'0 X~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ error: no match found same:16'1 ? possible intended match 70: !11 = !DILocation(line: 1, scope: !5) same:16'0 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 71: !12 = !DILocalVariable(name: "y", arg: 2, scope: !5, file: !3, line: 1, type: !9) same:16'0 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 72: !13 = !DILocation(line: 13, column: 2, scope: !5) same:16'0 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 73: !14 = !DILocation(line: 13, column: 6, scope: !5) same:16'0 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 74: !15 = !DILocation(line: 13, column: 9, scope: !5) same:16'0 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ >>>>>> FAIL With 1.88: OK ``` <details>

scrabsha force-pushed the scrabsha/html-errors branch from 47fbb79 to b33ccc3 Compare September 7, 2021 08:38

Sasha Pourcelot and others added 9 commits September 7, 2021 19:13

wait, wtf

be3af9f

Fix the whole thing, add a comment too

7e718d5

Fix typo, remove old fixme

ae79e44

Restructure code a bit, avoid an allocation

de28762

Remove margins, make raw strings

512d0b9

Add test file for html output

88614ef

Enable UTF-8

e2e20b1

Use split_inclusive to eliminate some calls to write

6757dbd

Put the escaper in its own state machine

5f212de

myrrlyn reviewed Sep 15, 2021

View reviewed changes

Sasha Pourcelot added 2 commits September 16, 2021 19:23

Don't report an error if we succeeded in writing at something

6e27b89

Use the segment index instead of a first boolean

8e064c1

scrabsha pushed a commit that referenced this pull request Jul 15, 2025

Merge pull request #1 from sesgoe/ses/update-clippy-fix-13518

52aa383

fix: check against 'main' function name instead of entrypoint function

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

Scrabsha/html errors #1

Scrabsha/html errors #1

Uh oh!

scrabsha commented Sep 7, 2021

Uh oh!

myrrlyn Sep 15, 2021

Uh oh!

scrabsha Sep 16, 2021

Uh oh!

myrrlyn commented Sep 15, 2021

Uh oh!

scrabsha commented Sep 16, 2021

Uh oh!

Uh oh!

Scrabsha/html errors #1

Are you sure you want to change the base?

Scrabsha/html errors #1

Uh oh!

Conversation

scrabsha commented Sep 7, 2021

Uh oh!

myrrlyn Sep 15, 2021

Choose a reason for hiding this comment

Uh oh!

scrabsha Sep 16, 2021

Choose a reason for hiding this comment

Uh oh!

myrrlyn commented Sep 15, 2021

Uh oh!

scrabsha commented Sep 16, 2021

Uh oh!

Uh oh!