Update IA 0.22 #345
Update IA 0.22 #345
Conversation
lbenet
force-pushed
the
lb/IA_1.0-dev
branch
from
02c6abb to
d03d017
Compare
lbenet
changed the title
lbenet
force-pushed
the
lb/IA_1.0-dev
branch
from
509d9c5 to
6a3be15
Compare
|
Tests are broken in Julia v1.6 because IA imposes Julia v1.9... |
|
Due to IntervalArithmetic requiring v1.9 of julia, the tests fail for Julia v1.6... |
lbenet
force-pushed
the
lb/IA_1.0-dev
branch
from
cf455d8 to
06dd3e8
Compare
lbenet
force-pushed
the
lb/IA_1.0-dev
branch
2 times, most recently
from
1955f41 to
ff9c3aa
Compare
|
Out of curiosity, what is the status of this PR? Are there any unresolved issues? |
|
Thanks @OlivierHnt for asking; so far so good (in terms of unresolved issues), though I haven't checked how things work after JuliaIntervals/IntervalArithmetic.jl#686. Things to do here are related to the ambiguities that generate IntervalArithmetic that affect the performance. (That's part of the reason I started to work on JuliaIntervals/IntervalArithmetic.jl#687.) I am also rewriting in #369 some parts related to It's been a quite stressfull semester, which has slowed down some of my efforts here (and in IntervalArithmetic); I hope to have more time in November... |
|
Sounds great :) |
|
@lrnv This branch upgrades what a bunch of things related to the use of |
|
@PerezHz There seems to be issues for JLD2 in Julia nightly; any ideas? |
|
@lbenet Thanks for pinging me there ! The package I use TaylorSeries in are WilliamsonTransforms.jl and Copulas.jl. If you want to check their tests yourself you can of course ! Otherwise please do what you have to do, publish your minor version, CompatHelper will do its job and if i have issues I'll come back to you. |
|
@PerezHz There seems to be issues for JLD2 in Julia nightly; any ideas? I tried reproducing the error on current nightly (1.13.0-DEV.284, macOS) but didn't manage (tests pass). The error points to JLD2.jl erroring during compilation which could point to an error from JLD2.jl or the specific nightly. Have you already tried simply rerunning? |
|
The failing tests on nightly used JLD2 v0.5.11; I tested locally with JLD2 v0.5.12 which was released yesterday including a fix for the errors in nightly, so tests should pass now there (if it manages to catch the new JLD2 version, which it should). |
I just checked the tests, and you are right. So I will relauch the tests. Thanks @PerezHz for checking it! |
JuliaDiff/TaylorSeries.jl#345 will add support to a newer version of IntervalArithmetics. This is needed to support recent versions of Makie. JuliaDiff/TaylorSeries.jl#345 bumps the version of TaylorSeries to 0.19, and this commit expands compat for that. CubedSphere.jl does not use IntervalArithmetics, so it is still compatible with 0.19. I tested 345 locally and the test suite for CubedSphere passes. (We can merge this after TaylorSeries 0.19 is released)
Very busy week(s), sorry for the silence @lrnv ... I've checked that Yet, I notice that |
|
I've checked that some packages that depend on TaylorSeries have passing tests with this PR. These packages include TaylorIntegration, PlanetaryEphemeris and NEOs; yet, TaylorModels depends on both, TaylorSeries and IntervalArithmetic, so it has to be considered separately. I'm thus merging this PR and tagging a new version. |
I guess that this is not as issue, since for someone using an older version of Julia, an older version of TaylorSeries.jl would be used right ? Anyway, thanks for checking, for the moment I ran tests and merged the compat PRs so we are good. Ty ! |
No description provided.