Backports to 1.0.1 #28764
Backports to 1.0.1 #28764
Conversation
KristofferC
force-pushed
the
kc/backports_1.0.1
branch
from
a90460f
to
227cefd
Compare
|
This is super cool! Excellent work as usual, Kristoffer! |
|
I've looked through the changes and it all seems good to me, but it'd be nice to have some more eyes on it. I think we should call this good for 1.0.1 and put remaining backports into a 1.0.2. |
|
We need #28822 too, includes lots of Pkg bugfixes. |
|
I made two small non-PR commits yesterday (doc typos) without thinking about backport, so I now just pushed them on this branch, hope it's ok. |
KristofferC
force-pushed
the
kc/backports_1.0.1
branch
from
6ba7876
to
c0a4b51
Compare
(cherry picked from commit 53d2e7c)
* Update documented REPL banner for 1.0 (cherry picked from commit 7c9383f)
(cherry picked from commit 7cb8531)
(cherry picked from commit 04f4884)
(cherry picked from commit 0846eb5)
* Doc IndexLinear and IndexCartesian (cherry picked from commit d15b091)
Add a punctuation (cherry picked from commit 883c8a3)
(cherry picked from commit decee65)
(cherry picked from commit 2fcad41)
(cherry picked from commit 3d8091f)
there is no real advantage in specializing on the argument types
for display since it goes through a quite complicated machinery
of trying to find applicable displays.
Before:
julia> let
@time precompile(Tuple{typeof(Base.Multimedia.display), Int32})
@time precompile(Tuple{typeof(Base.Multimedia.display), Vector{Int}})
@time precompile(Tuple{typeof(Base.Multimedia.display), Float64})
@time precompile(Tuple{typeof(Base.Multimedia.display), Symbol})
end
0.034542 seconds (37.08 k allocations: 1.916 MiB)
0.042272 seconds (92.57 k allocations: 4.810 MiB)
0.039003 seconds (90.20 k allocations: 4.758 MiB)
0.030826 seconds (61.08 k allocations: 3.066 MiB)
After:
julia> let
@time precompile(Tuple{typeof(Base.Multimedia.display), Int32})
@time precompile(Tuple{typeof(Base.Multimedia.display), Vector{Int}})
@time precompile(Tuple{typeof(Base.Multimedia.display), Float64})
@time precompile(Tuple{typeof(Base.Multimedia.display), Symbol})
end
0.000041 seconds (12 allocations: 640 bytes)
0.000029 seconds (10 allocations: 544 bytes)
0.000024 seconds (9 allocations: 496 bytes)
0.000021 seconds (9 allocations: 496 bytes)
(cherry picked from commit eabb601)
To get out of REPL. (cherry picked from commit 548d83f)
It was once an argument, but it's calculated on the first line, so... (cherry picked from commit bcce68d)
(cherry picked from commit 4e098f1)
(cherry picked from commit 7ffe704)
* improve performance for number to string conversion functions (cherry picked from commit 472fe5f)
(cherry picked from commit 290684d)
(cherry picked from commit 630f53a)
|
NewPkgEval showed regressions in the following packages:
So I think we're good! We just need to add #29354 to the backports here. |
|
I noticed that our FreeBSD CI is down. I can run the tests manually on FreeBSD tomorrow to ensure we haven't affected that at all, though I suspect it'll be fine. Worth checking for a point release though. |
|
I'm getting this on FreeBSD: It also happens on the |
|
Unless @JeffBezanson or @vtjnash can fix this on freebsd looking at that log, I think we should release and perhaps fix freebsd in a later point release. |
|
Or bisect to figure out which commit broke FreeBSD and then just drop that commit? |
|
|
Yeah I'd be fine releasing as-is given that it also fails on |
|
Fails on a build of the 1.0 tag. I think we're fine. |
unless the user has explicitly asked for it with --startup-file=yes (cherry picked from commit 40d7f27f2ff08ec466df536f267129a9f5e950b4) (cherry picked from commit eb96811)
(cherry picked from commit 57f7380a2641944be12695e92a3ad9f4cc20e6f2) (cherry picked from commit 5829a40)
The tests use InteractiveUtils but it is not listed as a test dependency in the SparseArrays Project.toml file. This is causing SparseArrays to fail its tests when run on NewPkgEval, which means every package that uses SparseArrays gets their tests skipped. (cherry picked from commit f2ad009)
(cherry picked from commit d974f93)
|
I plan to merge and push forward with a release once CI passes. |
|
|
Yeah that one has to be backported as well... |
(cherry picked from commit f8b52da)
|
#29315 backported to fix that CI error |
|
WHO LIKES RELEASES BECAUSE I DO |
|
Can someone help with removing some backport labels now... Or make a script that does it ;) |
|
Does this page help: https://github.com/JuliaLang/julia/labels/backport%20pending%201.0 ? |
|
As long as people make sure that the PR actually got backported. But if the script I used is not buggy, they all should have been... |
|
You have a list at the top of the PR of things marked for backport that will need to be backported manually I assume due to merge conflicts. We'll just have to make sure those retain their label. |
|
All of those have been backported already. Where do you see that list? |
|
"Non-merged PRs with backport label" section in the very top post of the PR. |
|
Oh, I assumed that was "non-merged" as in "not merged into this branch," i.e. backported. Derp. Don't mind me. |
?at end of input in conditional expression #28592 - fix Handling of a newline in ternary operator in REPL #28576, parsing?at end of input in conditional expressionprintstyledcalls #28586 - fix variousprintstyledcallsdisplay#28616 - stop specializing on argument types ofdisplay<:in typejoin and tuplemerge #28655 - fix Internal error in runtime when testing Meshing.jl #28641, passing typevars to<:in typejoin and tuplemergecluster_cookiebefore cluster starts #28656 - fixcluster_cookie()issuesUndefRefErroronv0.7-rc2#28499, callingcluster_cookiebefore cluster startsjl_value_tin h2 #28702 - Use code font forjl_value_tin h2reloadfrom the docs; fix #28572 #28644 - Remove mention ofreloadfrom the docs; fix Docs still mentionreload(...)#28572versioninfoin devdocs #28810 - Update info aboutversioninfoin devdocsRegexdocumentation #28703 - addRegexdocumentation?@specialize#28959 - fix typo in?@specializeSys.which()#28157mynormshould not divide by n #29091 - Documentationmynormshould not divide by ngetfield_nothrowto allow removing more getfields #29084 - fix logic ingetfield_nothrowto allow removing more getfieldssize#29094 - doc/interface: correct signature ofsizesshparameter in Parallel Computing manual #29101Iterators.flattenfor empty tuple #29112 - AllowIterators.flattenfor empty tuple@edit a[1]no longer works #25474, support more forms in edit, which, etc. macrosundefto Vector allocations #29184 - Addundefto Vector allocationsnew{}with incomplete type #29154 - fix Invalid syntax cause a crash in Julia 1.0 #29145, error fornew{}with incomplete typeNon-merged PRs with backport label:
evalin__init__#29134 - fix eval in __init__ causes incorrect precompilation warning #29059, unnecessary warning in precompile forevalin__init__CMD-object customization #28466 - DocumentCMD-object customization