Backports for 1.5.2 #37210

KristofferC · 2020-08-26T10:14:43Z

Backported PRs:

Contains multiple commits, manual intervention needed:

code coverage fixes #37243 - code coverage fixes

Non-merged PRs with backport label:

Fix function call to rec_backtrace_ctx when using USE_SYSTEM_LIBUNWIND=1 #36821 - Fix function call to rec_backtrace_ctx when using USE_SYSTEM_LIBUNWIND=1

(cherry picked from commit 9ae0a73)

Removes ambiguity: ``` julia> Base.OrderStyle(Union{}) ERROR: MethodError: Base.OrderStyle(::Type{Union{}}) is ambiguous. Candidates: ``` This error is relevant as with current `unique!` definition that relies on `OrderStyle` one can have a problem in corner cases. E.g.: ``` julia> [i for i in ["1"] if i isa Int] 0-element Array{Union{},1} ``` Co-authored-by: Milan Bouchet-Valat <[email protected]> (cherry picked from commit 03e1a89)

yuyichao · 2020-08-26T13:08:14Z

Seems that the back point of #37149 included more than just the bug fix. Only e change to gcutils.jl and secretbuffer.jl should be backported since those fixed the wrong return type from strlen call...

KristofferC · 2020-08-26T13:09:31Z

Thanks, I missed that comment, will change it.

…..." strings, fixes #16194. (#37130) (cherry picked from commit d446c27)

(cherry picked from commit 646626e)

We had been assuming that IdentityUnitRange matched the indices of the parent (like Slices) but they define their own offset axes. Further, other `AbstractRanges` may be similarly offset. We do not need to consider other offset `AbstractArrays` as this code only applies to `IndexLinear` `SubArray`s. (cherry picked from commit 701885b)

martinholters

The tests from #37443 need a minor adjustment.

test/compiler/inference.jl

JeffBezanson · 2020-09-09T05:01:17Z

There is a test failure here from 38c3e74. It looks like some part of dcc0696 is also needed to make that test work. That whole commit seems likely safe to backport though.

JeffBezanson · 2020-09-09T05:28:04Z

This is the needed line:

--- a/base/compiler/inferencestate.jl
+++ b/base/compiler/inferencestate.jl
@@ -205,7 +205,7 @@ update_valid_age!(edge::InferenceState, sv::InferenceState) = update_valid_age!(
 function record_ssa_assign(ssa_id::Int, @nospecialize(new), frame::InferenceState)
     old = frame.src.ssavaluetypes[ssa_id]
     if old === NOT_FOUND || !(new ⊑ old)
-        frame.src.ssavaluetypes[ssa_id] = tmerge(old, new)
+        frame.src.ssavaluetypes[ssa_id] = old === NOT_FOUND ? new : tmerge(old, new)
         W = frame.ip
         s = frame.stmt_types

martinholters · 2020-09-09T05:54:49Z

Ref. #37365 (comment) 😉

KristofferC · 2020-09-09T06:08:54Z

Strange, I was sure I did that...

KristofferC · 2020-09-09T08:24:39Z

@nanosoldier runbenchmarks(ALL, vs = ":release-1.5")

KristofferC · 2020-09-09T08:24:49Z

@nanosoldier runtests(ALL, vs = ":release-1.5")

nanosoldier · 2020-09-09T16:02:33Z

Your benchmark job has completed - possible performance regressions were detected. A full report can be found here. cc @ararslan

nanosoldier · 2020-09-10T03:19:12Z

Your package evaluation job has completed - possible new issues were detected. A full report can be found here. cc @maleadt

KristofferC · 2020-09-10T07:01:01Z

https://github.com/JuliaCI/NanosoldierReports/blob/801671c4f9782fadd4f9b6718570c91a1414fd0e/pkgeval/by_hash/21d877d_vs_c4acbf9/logs/StatProfilerHTML/1.5.2-pre-be8475f41a.log

ERROR: LoadError: LoadError: syntax: invalid assignment location "#40#content..." around /home/pkgeval/.julia/packages/HAML/hMyIa/src/Templates.jl:49
Stacktrace:
 [1] top-level scope at none:1
 [2] eval at ./boot.jl:331 [inlined]
 [3] eval at ./Base.jl:39 [inlined]
 [4] _includehaml(::Module, ::Symbol, ::String, ::String) at /home/pkgeval/.julia/packages/HAML/hMyIa/src/Templates.jl:98
 [5] includehaml at /home/pkgeval/.julia/packages/HAML/hMyIa/src/Templates.jl:68 [inlined] (repeats 2 times)
 [6] #3 at /home/pkgeval/.julia/packages/HAML/hMyIa/src/Templates.jl:71 [inlined]
 [7] foreach at ./abstractarray.jl:2009 [inlined]
 [8] includehaml(::Module, ::Pair{Symbol,String}, ::Pair{Symbol,String}, ::Pair{Symbol,String}, ::Pair{Symbol,String}, ::Vararg{Pair{Symbol,String},N} where N) at /home/pkgeval/.julia/packages/HAML/hMyIa/src/Templates.jl:70
 [9] top-level scope at /home/pkgeval/.julia/packages/StatProfilerHTML/BuyY0/src/HTML.jl:12
 [10] include(::Function, ::Module, ::String) at ./Base.jl:380
 [11] include at ./Base.jl:368 [inlined]
 [12] include(::String) at /home/pkgeval/.julia/packages/StatProfilerHTML/BuyY0/src/StatProfilerHTML.jl:1
 [13] top-level scope at /home/pkgeval/.julia/packages/StatProfilerHTML/BuyY0/src/StatProfilerHTML.jl:6
 [14] include(::Function, ::Module, ::String) at ./Base.jl:380

looks a bit strange. @JeffBezanson?

fredrikekre · 2020-09-10T08:12:53Z

Backport of #37315 should be HOME_PROJECT on 1.5 (#37315 (review))

(cherry picked from commit f780151)

(cherry picked from commit 3453a5c)

(cherry picked from commit f51d4cb)

(cherry picked from commit 405cd6f)

(cherry picked from commit 483b637)

While in 1d08d70 we wanted to avoid combining two types, we did not intend to prevent a type from being just a pass-through value. (cherry picked from commit a9743d8)

…terals (#37308) (cherry picked from commit f896006)

(cherry picked from commit 315ad46)

Fixes a crash when one of the incoming value for a union value phi node is of type `Union{}`. Incoming value of this type can be generated by try-catch. Fix #37265 (cherry picked from commit c3a63fc)

Unlike other nodes, Upsilon nodes can be potentially undefined and need special handling to ensure they aren't literally undef, as we might try to union-copy or gc-root the data later. Fix #37262 (cherry picked from commit 5cf5717)

This is a compile error on 10+ and is probably triggering implicity conversion on LLVM 8 and 9. (cherry picked from commit 5acd7bd)

The field should be either `Symbol` or `Int`. Ref #37423 Unlike #37423, in additional to worse type info in inference, the missing type check here can actually cause type inference error due to errors in user code. (cherry picked from commit 1378bb6)

Fixes #37180. (cherry picked from commit f047d7f)

In particular, make `obviously_unequal(Type{Union{}}, Type{T} where {Union{}<:T<:Union{}})` false. Fixes #37255 (cherry picked from commit 79a9ea0)

(cherry picked from commit 1e6d771)

(cherry picked from commit b17cd10)

(cherry picked from commit a83166a)

The undef ref error may be thrown at the wrong time or not at all and may cause crashes. (cherry picked from commit 72854dc)

KristofferC · 2020-09-15T07:37:50Z

This should be good to go once CI finishes.

JeffBezanson and others added 2 commits August 26, 2020 12:00

fix #37126, spurious soft scope warning on hidden name (#37178)

7764caf

(cherry picked from commit 9ae0a73)

yuyichao and others added 2 commits August 26, 2020 15:13

fix ccall of strlen

2e80350

Fix string interpolation into Markdown.Table and Markdown.List in md"…

c7dfcf6

…..." strings, fixes #16194. (#37130) (cherry picked from commit d446c27)

KristofferC force-pushed the backports-release-1.5 branch from f59a4ff to 359a71b Compare August 26, 2020 13:16

remove the incorrect return value attribute

96cfb7e

KristofferC force-pushed the backports-release-1.5 branch from 359a71b to 96cfb7e Compare August 26, 2020 13:17

fix #37134, error with macro returning x::T -> y (#37181)

2fef8fc

(cherry picked from commit 646626e)

greimel mentioned this pull request Aug 31, 2020

Wrap hasmethod(isless) in try ... catch block, check OrderStyle first JuliaData/CategoricalArrays.jl#292

Merged

JeffBezanson and others added 2 commits September 6, 2020 16:23

fix a possible segfault during static compilation (#37386)

764df9d

KristofferC force-pushed the backports-release-1.5 branch 3 times, most recently from 2bfbb88 to 4ea0e8c Compare September 8, 2020 12:22

martinholters reviewed Sep 8, 2020

View reviewed changes

test/compiler/inference.jl Outdated Show resolved Hide resolved

test/compiler/inference.jl Outdated Show resolved Hide resolved

KristofferC force-pushed the backports-release-1.5 branch from f529e8a to bb3ff89 Compare September 8, 2020 14:13

JeffBezanson force-pushed the backports-release-1.5 branch from 30d5003 to 332af19 Compare September 9, 2020 02:42

KristofferC force-pushed the backports-release-1.5 branch from e443b33 to 4a22a51 Compare September 9, 2020 06:36

KristofferC and others added 17 commits September 10, 2020 10:29

reset ACTIVE_PROJECT when building docs (#37315)

1efe7b7

(cherry picked from commit f780151)

Mark that ismutable requires julia 1.5 (#37357)

29a880b

(cherry picked from commit 3453a5c)

Make sure to build clangsa before running clangsa self tests (#37385)

3751599

(cherry picked from commit f51d4cb)

Allow packages to force depwarns (#37296)

1b1c443

(cherry picked from commit 405cd6f)

fix #37401, displaying Enum in the repl (#37404)

f5adf3d

(cherry picked from commit 483b637)

inference: fix precision regression (#37365)

dc57047

While in 1d08d70 we wanted to avoid combining two types, we did not intend to prevent a type from being just a pass-through value. (cherry picked from commit a9743d8)

fix #37282, printing juxtaposition of 0 confusion with hex,bin,oct li…

fc5df59

…terals (#37308) (cherry picked from commit f896006)

Build julia_version.h before running clangsa (#37394)

31f3c01

(cherry picked from commit 315ad46)

Only emit incoming values for the phi when needed (#37285)

d5b8d7e

Fixes a crash when one of the incoming value for a union value phi node is of type `Union{}`. Incoming value of this type can be generated by try-catch. Fix #37265 (cherry picked from commit c3a63fc)

Fix use of CreateAlignedStore (#37388)

23b061f

This is a compile error on 10+ and is probably triggering implicity conversion on LLVM 8 and 9. (cherry picked from commit 5acd7bd)

Prune egal types when forming a Union (#37207)

2961707

Fixes #37180. (cherry picked from commit f047d7f)

Fix type-equality involving Type{Union{}} (#37261)

5c96408

In particular, make `obviously_unequal(Type{Union{}}, Type{T} where {Union{}<:T<:Union{}})` false. Fixes #37255 (cherry picked from commit 79a9ea0)

add a part of dcc0696 needed to make tests pass

72d0e51

Specialize _memory_offset on number of arguments. (#37444)

bcc6a87

(cherry picked from commit 1e6d771)

Fix memory leak in umfpack_numeric! (#37472)

9f1d1fa

(cherry picked from commit b17cd10)

KristofferC force-pushed the backports-release-1.5 branch from 164abe0 to 9f1d1fa Compare September 10, 2020 08:29

JeffBezanson mentioned this pull request Sep 11, 2020

fix "invalid assignment location" macro hygiene regression #37534

Merged

JeffBezanson closed this in #37534 Sep 12, 2020

fredrikekre reopened this Sep 12, 2020

JeffBezanson and others added 3 commits September 15, 2020 09:27

fix "invalid assignment location" macro hygiene regression (#37534)

8cb458c

(cherry picked from commit a83166a)

Fix getfield of potentially undef inline immutable field (#37511)

765e1b3

The undef ref error may be thrown at the wrong time or not at all and may cause crashes. (cherry picked from commit 72854dc)

bump Pkg to 1.5.2

b8b7267

KristofferC added the release Release management and versioning. label Sep 15, 2020

KristofferC merged commit 005352a into release-1.5 Sep 15, 2020

KristofferC deleted the backports-release-1.5 branch September 15, 2020 15:03

KristofferC changed the title ~~WIP: Backports for 1.5.2~~ Backports for 1.5.2 Sep 16, 2020

johnnychen94 mentioned this pull request Sep 18, 2020

update compat patch version on #133 JuliaArrays/OffsetArrays.jl#152

Merged

Uh oh!

Backports for 1.5.2 #37210

Backports for 1.5.2 #37210

Uh oh!

Conversation

KristofferC commented Aug 26, 2020 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

yuyichao commented Aug 26, 2020

Uh oh!

KristofferC commented Aug 26, 2020

Uh oh!

martinholters left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

JeffBezanson commented Sep 9, 2020

Uh oh!

JeffBezanson commented Sep 9, 2020

Uh oh!

martinholters commented Sep 9, 2020

Uh oh!

KristofferC commented Sep 9, 2020

Uh oh!

KristofferC commented Sep 9, 2020

Uh oh!

KristofferC commented Sep 9, 2020

Uh oh!

nanosoldier commented Sep 9, 2020

Uh oh!

nanosoldier commented Sep 10, 2020

Uh oh!

KristofferC commented Sep 10, 2020

Uh oh!

fredrikekre commented Sep 10, 2020

Uh oh!

KristofferC commented Sep 15, 2020

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

16 participants

KristofferC commented Aug 26, 2020 •

edited

Loading