Fix world age problems when running tests #41

timholy · 2019-02-15T15:13:20Z

This WiP solves the world age problems by an inversion of control: rather than having lower_incrementally(f, mod, expr) recurse into expr, lower its individual pieces, and then running the resulting frames with f, this redesigns prepare_toplevel so that it splits expr into frames and returns it as a task list. The caller is then responsible for running each of them. While this is extra work for the caller, the big advantage is that if you do this in top-level you can ensure that there is a world age update after each new method definition. (This is achieved just by returning "prematurely" after each 3-arg :method expr is encountered.)

The core tests pass, but I'm having surprising difficulty rewriting juliatests.jl to use this. I may have to pivot to other things for the remainder of the day, so please don't merge yet. (I did want people to know this is coming---it's a pretty disruptive change.)

Bonus: includes new docs!

codecov-io · 2019-02-15T15:25:45Z

Codecov Report

Merging #41 into master will not change coverage.
The diff coverage is 0%.

@@          Coverage Diff          @@
##           master    #41   +/-   ##
=====================================
  Coverage       0%     0%           
=====================================
  Files           6      6           
  Lines        1115   1117    +2     
=====================================
- Misses       1115   1117    +2

Impacted Files	Coverage Δ
src/precompile.jl	`0% <ø> (ø)`	⬆️
src/JuliaInterpreter.jl	`0% <0%> (ø)`	⬆️
src/interpret.jl	`0% <0%> (ø)`	⬆️
src/builtins.jl	`0% <0%> (ø)`	⬆️

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 64cc800...26e04b9. Read the comment docs.

KristofferC · 2019-02-15T15:30:06Z

Is is true that theoretically, we could just invokelatest every function? In that case, perhaps it should be mentioned why we don't do that.

timholy · 2019-02-15T15:38:24Z

That was indeed the old bandaid, you can find such a call in lower!. But that became not-very-viable when I discovered that I'd also need it on calls in maybe_evaluate_builtin: it would make the interpreter insanely slow. Since this is an issue that affects the test suite much more than ordinary usage via @interpret foo(args...), I thought it was better to isolate the pain.

KristofferC · 2019-02-15T15:45:07Z

Yeah, I didn't mean to suggest it as a workaround but just having a note about it in the docs to immediately catch the people going "oh they don't know about invokelatest to get around world error"-people.

timholy · 2019-02-15T15:45:59Z

Makes total sense, great idea.

pfitzseb · 2019-02-15T16:09:01Z

Maybe I'm missing something, but wouldn't it be possible to insert a manual world age increment after :method expressions?

timholy · 2019-02-15T16:25:03Z

I don't know how to do that, do you?

pfitzseb · 2019-02-16T10:49:44Z

I don't know how to do that, do you?

I don't -- turns out I didn't actually understand the problem properly (I thought the global world age wasn't incremented properly, but the problem is that the "local" world age is constant despite a new method being defined). So yeah, I can't think of a better way to do this.

timholy · 2019-02-16T10:56:24Z

@pfitzseb, I have long suspected that the omission of manual world-age updates is deliberate, but I hadn't really thought carefully until prompted by your question. Now my guess is this: if Julia supported manual updates, what would happen for a compiled function that happened to redefine itself and then trigger a world age update? Or a function that redefined one of its callers?

In contrast, if world age only updates when you are at top level, there is no caller (well, there's potentially eval_user_input, but this is a special case thats easily caught). So that situation can never arise.

Note that base "cheats" a bit on this: see, for example, https://github.com/JuliaLang/julia/blob/68db87147f9806db605c4df9f32b0869554cf4b7/src/interpreter.c#L883-L885. Since we can't do that, we have to employ other strategies.

timholy · 2019-02-16T11:02:40Z

I now have more of an idea of why I'm struggling with this. Check out this bizarre Julia bug:

...
    frame.pc[] = pc
    # Handle the return
    stmt = pc_expr(frame, pc)
    if nstmts == 0 && !isexpr(stmt, :return)
        println("aborted")
        JuliaInterpreter.show_stackloc(stack, frame, pc)
        @show pc frame
        ret = Aborted(frame, pc)
        @show ret
        return ret, nstmts
    end
...

and that Aborted constructor looks like this

function Aborted(frame::JuliaStackFrame, pc)
    println("Aborted constructor")
    @show pc frame
    lineidx = frame.code.code.codelocs[convert(Int, pc)]
    @show lineidx
    return Aborted(frame.code.code.linetable[lineidx])
end

yet I'm seeing output like

aborted
Main.JuliaTests
  Main.JuliaTests
    showerror(io::IO, ex::MethodError) in Base at errorshow.jl:168
      showerror_ambiguous(io::IO, meth, f, args) in Base at errorshow.jl:271
        print(io::IO, xs...) in Base at strings/io.jl:39
          print(io::IO, x) in Base at strings/io.jl:29
            show(io::IO, m::Method) in Base at methodshow.jl:113
              #show#520(kwtype::Union{Nothing, DataType}, ::Any, io::IO, m::Method) in Base at methodshow.jl:113
                arg_decl_parts(m::Method) in Base at methodshow.jl:55
                  argtype_decl(env, n, sig::DataType, i::Int64, nargs, isva::Bool) in Base at methodshow.jl:6
                    string_with_env(env, xs...) in Base at strings/io.jl:129
                      print(io::IO, x) in Base at strings/io.jl:29
                        show(io::IO, x::DataType) in Base at show.jl:436
                          show_datatype(io::IO, x::DataType) in Base at show.jl:511
                            show_type_name(io::IO, tn::Core.TypeName) in Base at show.jl:450
                              print(io::IO, s::Union{SubString{String}, String}) in Base at strings/io.jl:167
                                write(io::IO, s::Union{SubString{String}, String}) in Base at strings/io.jl:165
                                  unsafe_write(io::Base.AbstractPipe, p::Ptr{UInt8}, nb::UInt64) in Base at io.jl:225
                                    unsafe_write(to::Base.GenericIOBuffer, p::Ptr{UInt8}, nb::UInt64) in Base at iobuffer.jl:409
                                      getproperty(x, f::Symbol) in Base at sysimg.jl:18, pc = 2
pc = JuliaProgramCounter(2)
frame = JuliaStackFrame(JuliaInterpreter.JuliaFrameCode(getproperty(x, f::Symbol) in Base at sysimg.jl:18, CodeInfo(
1 ─      nothing
│   %2 = ($(QuoteNode(getfield)))(x, f)
└──      return %2
), #= stuff =#)
Aborted constructor
pc = JuliaProgramCounter(140359322445712)   # what?!? wasn't it at 2 moments ago?
frame = JuliaStackFrame(JuliaInterpreter.JuliaFrameCode(getproperty(x, f::Symbol) in Base at sysimg.jl:18, CodeInfo(
1 ─      nothing
│   %2 = ($(QuoteNode(getfield)))(x, f)
└──      return %2
...

KristofferC · 2019-02-16T11:10:35Z

Yikes 😨

timholy · 2019-02-16T19:24:26Z

OK, this seems to work now, and adds in a fix for generated functions as a bonus. With this the tuple test finishes.

However, one negative is that this disables the parallel framework and runs all tests in id 1. Because of the tests that kill Julia, this means we don't finish them. I haven't yet looked into how to run these tests in top level on the workers (all the @async stuff goes through thunks, AFAIU). Maybe I need to eval them on the workers? Thoughts?

I'll point out another oddity in the code.

timholy · 2019-02-16T19:25:22Z

test/toplevel.jl

-    @test @interpret(isdefined(Toplevel, :Beat))
-    @test @interpret(Toplevel.Beat <: Toplevel.DatesMod.Period)
+    # @test @interpret(isdefined(Toplevel, :Beat))
+    # @test @interpret(Toplevel.Beat <: Toplevel.DatesMod.Period)


Bizarrely, these seem to break the printing of the TestSet results. I haven't looked into the cause.

timholy · 2019-02-16T19:28:08Z

Dang, passed on a 2-day old master. I'll update and see what might need fixing.

timholy · 2019-02-16T23:33:03Z

Here's an update on the tests with this branch (I had to comment out the `@sync` wrapper and put in a `sleep(300)` to let still-running tests finish). It's not an unequivocal win, there are even some regressions. More work to do, I fear 😢.

Julia Version 1.2.0-DEV.344 Commit a44b421138* (2019-02-16 16:52 UTC) Platform Info: OS: Linux (x86_64-linux-gnu) CPU: Intel(R) Xeon(R) CPU E5-2640 v3 @ 2.60GHz WORD_SIZE: 64 LIBM: libopenlibm LLVM: libLLVM-6.0.1 (ORCJIT, haswell) Environment: JULIA_CPU_THREADS = 8 Test run at: 2019-02-16T17:24:21.508

Maximum number of statements per lowered expression: 1000000

Test file	Passes	Fails	Errors	Broken	Aborted blocks
ambiguous	62	0	0	2	0
subarray	278	0	1	0	1
strings/basic	7211	1	4	0	3
strings/search	522	0	0	0	0
strings/util	449	0	0	0	0
strings/io	40	0	0	0	1
strings/types	X	X	X	X	X
unicode/utf8	19	0	0	0	0
core	X	X	X	X	X
keywordargs	X	X	X	X	X
numbers	X	X	X	X	X
subtype	X	X	X	X	X
char	☠️	☠️	☠️	☠️	☠️
triplequote	28	0	0	0	0
intrinsics	44	0	0	0	0
dict	X	X	X	X	X
hashing	X	X	X	X	X
iobuffer	151	0	0	0	1
staged	X	X	X	X	X
offsetarray	X	X	X	X	X
arrayops	X	X	X	X	X
tuple	488	0	1	0	0
reduce	1413	0	3	0	4
reducedim	X	X	X	X	X
abstractarray	X	X	X	X	X
intfuncs	4364	0	15	0	1
simdloop	61	3	0	0	1
vecelement	X	X	X	X	X
rational	X	X	X	X	X
bitarray	X	X	X	X	X
copy	493	11	7	0	0
math	X	X	X	X	X
fastmath	907	3	3	0	0
functional	X	X	X	X	X
iterators	601	0	4	0	2
operators	3758	0	0	0	1
path	X	X	X	X	X
ccall	X	X	X	X	X
parse	X	X	X	X	X
loading	X	X	X	X	X
bigint	X	X	X	X	X
sorting	985	0	65	0	4
spawn	X	X	X	X	X
backtrace	X	X	X	X	X
exceptions	28	19	6	0	0
file	X	X	X	X	X
read	X	X	X	X	X
version	1134	0	0	0	1
namedtuple	161	0	8	1	0
mpfr	758	0	45	0	0
broadcast	X	X	X	X	X
complex	X	X	X	X	X
floatapprox	49	0	0	0	0
reflection	X	X	X	X	X
regex	38	0	1	0	0
float16	124	0	0	0	0
combinatorics	92	0	3	0	1
sysinfo	2	0	0	0	0
env	X	X	X	X	X
rounding	X	X	X	X	X
ranges	X	X	X	X	X
mod2pi	80	0	0	0	0
euler	4	0	4	0	4
show	X	X	X	X	X
client	0	3	0	0	0
errorshow	X	X	X	X	X
sets	X	X	X	X	X
goto	X	X	X	X	X
llvmcall	X	X	X	X	X
llvmcall2	6	0	0	0	0
grisu	683	0	0	0	1
some	64	0	0	0	0
meta	X	X	X	X	X
stacktraces	X	X	X	X	X
docs	X	X	X	X	X
misc	X	X	X	X	X
threads	X	X	X	X	X
enums	X	X	X	X	X
cmdlineargs	X	X	X	X	X
int	X	X	X	X	X
checked	X	X	X	X	X
bitset	192	0	0	0	0
floatfuncs	68	0	1	0	1
boundscheck	X	X	X	X	X
error	31	0	0	0	0
cartesian	7	0	0	0	0
osutils	42	0	0	0	0
channels	X	X	X	X	X
iostream	11	0	2	0	0
secretbuffer	17	0	0	0	0
specificity	X	X	X	X	X
reinterpretarray	91	0	0	0	1
syntax	X	X	X	X	X
logging	X	X	X	X	X
missing	X	X	X	X	X
asyncmap	X	X	X	X	X
SparseArrays/higherorderfns	X	X	X	X	X
SparseArrays/sparse	X	X	X	X	X
SparseArrays/sparsevector	X	X	X	X	X
Pkg/resolve	X	X	X	X	X
LinearAlgebra/triangular	X	X	X	X	X
LinearAlgebra/qr	X	X	X	X	X
LinearAlgebra/dense	X	X	X	X	X
LinearAlgebra/matmul	X	X	X	X	X
LinearAlgebra/schur	X	X	X	X	X
LinearAlgebra/special	X	X	X	X	X
LinearAlgebra/eigen	X	X	X	X	X
LinearAlgebra/bunchkaufman	X	X	X	X	X
LinearAlgebra/svd	X	X	X	X	X
LinearAlgebra/lapack	X	X	X	X	X
LinearAlgebra/tridiag	X	X	X	X	X
LinearAlgebra/bidiag	X	X	X	X	X
LinearAlgebra/diagonal	X	X	X	X	X
LinearAlgebra/cholesky	X	X	X	X	X
LinearAlgebra/lu	X	X	X	X	X
LinearAlgebra/symmetric	X	X	X	X	X
LinearAlgebra/generic	X	X	X	X	X
LinearAlgebra/uniformscaling	X	X	X	X	X
LinearAlgebra/lq	X	X	X	X	X
LinearAlgebra/hessenberg	X	X	X	X	X
LinearAlgebra/blas	X	X	X	X	X
LinearAlgebra/adjtrans	X	X	X	X	X
LinearAlgebra/pinv	X	X	X	X	X
LinearAlgebra/givens	X	X	X	X	X
LinearAlgebra/structuredbroadcast	X	X	X	X	X
LibGit2/libgit2	X	X	X	X	X
Dates/accessors	X	X	X	X	X
Dates/adjusters	X	X	X	X	X
Dates/query	X	X	X	X	X
Dates/periods	X	X	X	X	X
Dates/ranges	X	X	X	X	X
Dates/rounding	X	X	X	X	X
Dates/types	X	X	X	X	X
Dates/io	X	X	X	X	X
Dates/arithmetic	X	X	X	X	X
Dates/conversions	X	X	X	X	X
Base64	20	0	0	0	1
CRC32c	658	0	6	0	0
DelimitedFiles	X	X	X	X	X
FileWatching	X	X	X	X	X
Future	0	0	0	0	0
InteractiveUtils	X	X	X	X	X
Libdl	X	X	X	X	X
Logging	33	1	2	0	0
Markdown	X	X	X	X	X
Mmap	X	X	X	X	X
Printf	X	X	X	X	X
Profile	X	X	X	X	X
REPL	995	0	0	5	0
Random	X	X	X	X	X
SHA	X	X	X	X	X
Serialization	X	X	X	X	X
Sockets	X	X	X	X	X
Statistics	X	X	X	X	X
SuiteSparse	778	0	0	0	0
Test	X	X	X	X	X
UUIDs	22	0	0	0	0
Unicode	☠️	☠️	☠️	☠️	☠️

timholy · 2019-02-17T15:19:45Z

OK, I think this is finally ready to merge. It's failing on nightly because of Debugger's tests; however, presumably Debugger won't pass on nightly without this (specifically, b1f7906). I won't merge right away in case @KristofferC wants to use this as a test case for https://github.com/JuliaLang/Pkg.jl/issues/770. (Unless committing the manifest is "the answer," in which case perhaps that can be closed.)

Among many other changes, this now has an interesting "fallback to Compiled" behavior here. This makes sure a test runs even if the interpreter gives up on it. Consequently we don't get failures like #30, You can still have problems if a test has done some work (e.g., printing to a buffer) but not finished; when you re-run it under Compiled that work will be repeated (which can cause some tests to fail). So this doesn't perfectly eliminate spurious faiilures, but it's getting closer. See #44 for discussion.

As a consequence, the tests take much longer to run than they used to.

Fixes #36. Fixes #30. Fixes #41

It's now a heck of a lot better, though clearly there's still some ways to go. Expand for results. However, I'm not entirely certain this is correct; for example, if I run them one-by-one I get the following for a selection of the tests that report as X in the big run below:

Test file	Passes	Fails	Errors	Aborted blocks
offsetarray	352	0	0	3
reducedim	686	3	0	27
bigint	2155	0	1	18
sorting	4864	0	0	8

so there may be more passing than this suggests. Needs investigation.

Julia Version 1.2.0-DEV.344
Commit a44b421138* (2019-02-16 16:52 UTC)
Platform Info:
OS: Linux (x86_64-linux-gnu)
CPU: Intel(R) Xeon(R) CPU E5-2640 v3 @ 2.60GHz
WORD_SIZE: 64
LIBM: libopenlibm
LLVM: libLLVM-6.0.1 (ORCJIT, haswell)
Environment:
JULIA_CPU_THREADS = 8
Test run at: 2019-02-17T09:17:51.447

Maximum number of statements per lowered expression: 1000000

Test file	Passes	Fails	Errors	Broken	Aborted blocks
ambiguous	62	0	0	2	0
subarray	280	0	1	0	1
strings/basic	87290	1	4	0	3
strings/search	522	0	0	0	0
strings/util	449	0	0	0	0
strings/io	12749	0	0	0	1
strings/types	2302685	0	3	0	4
unicode/utf8	19	0	0	0	0
core	X	X	X	X	X
worlds	☠️	☠️	☠️	☠️	☠️
keywordargs	133	0	1	0	0
numbers	1518443	27	284	0	6
subtype	X	X	X	X	X
char	☠️	☠️	☠️	☠️	☠️
triplequote	28	0	0	0	0
intrinsics	44	0	0	0	0
dict	144290	1	1	0	8
hashing	X	X	X	X	X
iobuffer	200	0	0	0	1
staged	57	3	0	0	0
offsetarray	X	X	X	X	X
arrayops	☠️	☠️	☠️	☠️	☠️
tuple	488	0	1	0	0
reduce	8465	0	3	0	4
reducedim	X	X	X	X	X
abstractarray	☠️	☠️	☠️	☠️	☠️
intfuncs	4404	0	15	0	1
simdloop	X	X	X	X	X
vecelement	X	X	X	X	X
rational	97493	0	29	0	2
bitarray	899682	0	0	0	9
copy	493	11	7	0	0
math	1505698	0	48	0	5
fastmath	907	3	3	0	0
functional	95	0	0	0	0
iterators	1568	0	4	0	2
operators	12928	0	0	0	1
path	274	0	0	12	2
ccall	X	X	X	X	X
parse	10281	20	1	0	1
loading	143633	289	1	0	10
bigint	X	X	X	X	X
sorting	X	X	X	X	X
spawn	X	X	X	X	X
backtrace	5	9	12	1	0
exceptions	28	19	6	0	0
file	X	X	X	X	X
read	X	X	X	X	X
version	2468	0	0	0	1
namedtuple	161	0	8	1	0
mpfr	758	0	45	0	0
broadcast	416	0	7	0	2
complex	8249	0	1	2	1
floatapprox	49	0	0	0	0
reflection	X	X	X	X	X
regex	38	0	1	0	0
float16	124	0	0	0	0
combinatorics	95	0	3	0	1
sysinfo	2	0	0	0	0
env	X	X	X	X	X
rounding	112496	0	1	0	2
ranges	12108707	2	209	327749	7
mod2pi	80	0	0	0	0
euler	8	0	4	0	4
show	X	X	X	X	X
client	0	3	0	0	0
errorshow	X	X	X	X	X
sets	739	0	1	0	1
goto	X	X	X	X	X
llvmcall	X	X	X	X	X
llvmcall2	6	0	0	0	0
grisu	X	X	X	X	X
some	64	0	0	0	0
meta	X	X	X	X	X
stacktraces	X	X	X	X	X
docs	X	X	X	X	X
misc	X	X	X	X	X
threads	X	X	X	X	X
enums	84	3	3	0	0
cmdlineargs	X	X	X	X	X
int	10545	0	184	0	1
checked	1102	0	9	0	0
bitset	192	0	0	0	0
floatfuncs	116	0	1	0	1
boundscheck	X	X	X	X	X
error	31	0	0	0	0
cartesian	7	0	0	0	0
osutils	42	0	0	0	0
channels	☠️	☠️	☠️	☠️	☠️
iostream	11	0	2	0	0
secretbuffer	17	0	0	0	0
specificity	X	X	X	X	X
reinterpretarray	128	0	0	0	1
syntax	X	X	X	X	X
logging	117	2	0	0	0
missing	406	0	0	1	1
asyncmap	X	X	X	X	X
SparseArrays/higherorderfns	7031	64	0	73	8
SparseArrays/sparse	☠️	☠️	☠️	☠️	☠️
SparseArrays/sparsevector	9884	0	1	0	5
Pkg/resolve	128	1	10	0	1
LinearAlgebra/triangular	33194	0	0	0	2
LinearAlgebra/qr	☠️	☠️	☠️	☠️	☠️
LinearAlgebra/dense	7692	0	14	0	7
LinearAlgebra/matmul	X	X	X	X	X
LinearAlgebra/schur	390	0	0	0	1
LinearAlgebra/special	2863	0	0	0	6
LinearAlgebra/eigen	406	0	0	0	2
LinearAlgebra/bunchkaufman	5145	0	0	0	1
LinearAlgebra/svd	X	X	X	X	X
LinearAlgebra/lapack	☠️	☠️	☠️	☠️	☠️
LinearAlgebra/tridiag	1222	0	0	0	2
LinearAlgebra/bidiag	1946	0	0	0	1
LinearAlgebra/diagonal	1619	0	0	0	2
LinearAlgebra/cholesky	2194	0	0	0	1
LinearAlgebra/lu	1182	0	3	0	1
LinearAlgebra/symmetric	2076	0	0	0	2
LinearAlgebra/generic	436	0	2	0	3
LinearAlgebra/uniformscaling	X	X	X	X	X
LinearAlgebra/lq	1253	0	0	0	1
LinearAlgebra/hessenberg	40	0	0	0	0
LinearAlgebra/blas	628	0	0	0	1
LinearAlgebra/adjtrans	253	0	0	0	1
LinearAlgebra/pinv	288	0	0	0	1
LinearAlgebra/givens	1840	0	0	0	1
LinearAlgebra/structuredbroadcast	408	0	0	0	2
LibGit2/libgit2	57	0	80	0	0
Dates/accessors	7723858	0	0	0	4
Dates/adjusters	3147	0	0	0	4
Dates/query	988	0	0	0	0
Dates/periods	679	0	2	0	1
Dates/ranges	349123	0	0	0	5
Dates/rounding	296	0	0	0	0
Dates/types	169	0	2	0	0
Dates/io	100	22	166	0	0
Dates/arithmetic	318	0	0	0	0
Dates/conversions	160	0	0	0	0
Base64	1015	0	0	0	1
CRC32c	658	0	6	0	0
DelimitedFiles	X	X	X	X	X
FileWatching	X	X	X	X	X
Future	0	0	0	0	0
InteractiveUtils	96	3	7	0	4
Libdl	X	X	X	X	X
Logging	35	1	0	0	1
Markdown	X	X	X	X	X
Mmap	☠️	☠️	☠️	☠️	☠️
Printf	701	38	0	0	0
Profile	15	0	0	0	2
REPL	995	0	0	5	0
Random	X	X	X	X	X
SHA	X	X	X	X	X
Serialization	X	X	X	X	X
Sockets	X	X	X	X	X
Statistics	606	0	1	0	4
SuiteSparse	778	0	0	0	0
Test	X	X	X	X	X
UUIDs	22	0	0	0	0
Unicode	☠️	☠️	☠️	☠️	☠️

Also fixes an issue in which next_until! failed to store the final pc.

This reworks prepare_toplevel so that it produces a list of frames that can be executed in order. In conjunction with another addition, through_methoddef_or_done!, one can run each frame sequentially up to the point of a new method definition, after which control can return to top-level. This allows the world age to update. While this places more burden on the top-level evaluator (one now requires a nest of two loops), this resolves the difficulties with previous strategies. It fixes a test previously marked as broken.

Also adds some line number info in prepare_toplevel to assist testing

This avoids the need for a macroexpand in prepare_toplevel, and therefore preserves the block organization of the original code. This also fixes a bug in optimize! which would inappropriately extract "inner methods" from :toplevel expressions.

Also fixes a bug when trying to run a :thunk expr in Compiled mode.

The problem here is that lowering can depend on the result of a previous frame. Consider an example in Julia's test/offsetarray.jl: stuff module NewModule using Test atsign-test_throws blah blah end That `atsign-test_throws` cannot be lowered until the `using Test` has been executed inside `NewModule`.

timholy · 2019-02-18T15:54:26Z

I will merge this once tests pass.

KristofferC · 2019-02-18T16:01:53Z

Restarted test to pickup Debugger fix

timholy · 2019-02-18T16:09:59Z

There will be an output difference in the Debugger tests, but let's deal with that separately.

https://travis-ci.org/JuliaDebug/JuliaInterpreter.jl/jobs/494967019

KristofferC · 2019-02-18T16:10:19Z

👍

timholy · 2019-02-18T16:22:12Z

I've updated the table in #13. Also check out the note about the crashes:

ones that caused errors or process exit (note: with the possible exceptions of channels, worlds, and arrayops, it appears that most such errors are due to a single cause, #28; deleting this block and rebuilding Julia fixes them)

timholy force-pushed the teh/world_age branch from 26e04b9 to 4e8e63d Compare February 16, 2019 18:47

timholy changed the title ~~WIP: fix world age problems when running tests~~ Fix world age problems when running tests Feb 16, 2019

timholy force-pushed the teh/world_age branch from 4e8e63d to 321fbf3 Compare February 16, 2019 19:21

timholy commented Feb 16, 2019

View reviewed changes

timholy mentioned this pull request Feb 17, 2019

Compiled resumers #44

Open

timholy added 10 commits February 18, 2019 09:53

Test for world age issues

2c9b2d4

Also fixes an issue in which next_until! failed to store the final pc.

Update the docs for API modifications, and describe top level issues

6e332fc

Replace limited_finish_and_return! with evaluate_limited!

2b12011

Also adds some line number info in prepare_toplevel to assist testing

Skip :method thunks

e6149a5

Generate correct error for generators that throw

cc971ed

prepare_toplevel: always store expression for simpler debugging

59f1cfb

Rewrite juliatests.jl to use new top-level API

d161dc5

Comment out tests that seem to break test printing

dcf573f

Fixes for very recent master

c8a0c0d

timholy added 9 commits February 18, 2019 09:53

Restore parallel juliatests

8f4b33e

Move julia test body to a function that calls Core.eval

7d8eaf2

Handle :toplevel expressions when stepping

e66a3f7

This avoids the need for a macroexpand in prepare_toplevel, and therefore preserves the block organization of the original code. This also fixes a bug in optimize! which would inappropriately extract "inner methods" from :toplevel expressions.

Finish aborted tests in Compiled mode. Fixes #30.

01e7f2a

Also fixes a bug when trying to run a :thunk expr in Compiled mode.

Circumvent worker-failure problems with a sleep

8ac9a73

Restore sync to juliatests, catch InterruptExceptions

70580fd

Better isolation of tests in juliatests

c497fee

Describe invokelatest in docs re world age

2c64810

timholy force-pushed the teh/world_age branch from f63f241 to 2c64810 Compare February 18, 2019 15:53

timholy mentioned this pull request Feb 18, 2019

Error in copy_codeinfo no method matching copy(::Type{Any}) in Debugger tests on nightly #46

Closed

timholy merged commit 7705307 into master Feb 18, 2019

timholy deleted the teh/world_age branch February 18, 2019 16:17

timholy mentioned this pull request Feb 18, 2019

MWE for InteractiveUtils fail #36

Closed

timholy mentioned this pull request Mar 5, 2019

MWE failure in SHA #101

Closed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Fix world age problems when running tests #41

Fix world age problems when running tests #41

timholy commented Feb 15, 2019 •

edited

Loading

codecov-io commented Feb 15, 2019 •

edited

Loading

KristofferC commented Feb 15, 2019

timholy commented Feb 15, 2019

KristofferC commented Feb 15, 2019

timholy commented Feb 15, 2019

pfitzseb commented Feb 15, 2019

timholy commented Feb 15, 2019

pfitzseb commented Feb 16, 2019

timholy commented Feb 16, 2019

timholy commented Feb 16, 2019

KristofferC commented Feb 16, 2019

timholy commented Feb 16, 2019

timholy Feb 16, 2019

timholy commented Feb 16, 2019

timholy commented Feb 16, 2019 •

edited

Loading

timholy commented Feb 17, 2019 •

edited

Loading

timholy commented Feb 18, 2019

KristofferC commented Feb 18, 2019

timholy commented Feb 18, 2019 •

edited

Loading

KristofferC commented Feb 18, 2019

timholy commented Feb 18, 2019

Fix world age problems when running tests #41

Fix world age problems when running tests #41

Conversation

timholy commented Feb 15, 2019 • edited Loading

codecov-io commented Feb 15, 2019 • edited Loading

Codecov Report

KristofferC commented Feb 15, 2019

timholy commented Feb 15, 2019

KristofferC commented Feb 15, 2019

timholy commented Feb 15, 2019

pfitzseb commented Feb 15, 2019

timholy commented Feb 15, 2019

pfitzseb commented Feb 16, 2019

timholy commented Feb 16, 2019

timholy commented Feb 16, 2019

KristofferC commented Feb 16, 2019

timholy commented Feb 16, 2019

timholy Feb 16, 2019

Choose a reason for hiding this comment

timholy commented Feb 16, 2019

timholy commented Feb 16, 2019 • edited Loading

timholy commented Feb 17, 2019 • edited Loading

timholy commented Feb 18, 2019

KristofferC commented Feb 18, 2019

timholy commented Feb 18, 2019 • edited Loading

KristofferC commented Feb 18, 2019

timholy commented Feb 18, 2019

timholy commented Feb 15, 2019 •

edited

Loading

codecov-io commented Feb 15, 2019 •

edited

Loading

timholy commented Feb 16, 2019 •

edited

Loading

timholy commented Feb 17, 2019 •

edited

Loading

timholy commented Feb 18, 2019 •

edited

Loading