Upgrade to Julia 0.7/1.0 #34
Conversation
| @@ -3,7 +3,6 @@ os: | |||
| - linux | |||
| - osx | |||
| julia: | |||
| - 0.6 | |||
| - 0.7 | |||
| - nightly | |||
src/expmv.jl
Outdated
| warn("opnorm($(typeof(A)), Inf) is not defined, fall back to using `anorm = 1.0`. | ||
| To suppress this warning, please specify the anorm parameter manually.") | ||
| @warn "opnorm($(typeof(A)), Inf) is not defined, fall back to using `anorm = 1.0`. | ||
| To suppress this warning, please specify the anorm keyword manually." |
There was a problem hiding this comment.
unrelated to 1.0 compatibility, but the second mention of anorm could have backquotes around it
| @test res < 1e-6 | ||
|
|
||
| #= commented since this test takes sometime difs very long |
There was a problem hiding this comment.
this is new. do these tests take longer on v0.7 than previously? what does difs mean?
There was a problem hiding this comment.
Sorry! Yes, this took much longer, but I didn't investigate what causes it. ("difs" is just a weird typo) Since the other tests for chbv are Ok, I took this one out.
There was a problem hiding this comment.
Everything generally looks good to me. I looked carefully at the main code and only skimmed the changes to the tests. I agree it makes most sense to drop support for julia 0.6. Many packages are doing this already, and this package is a special case as well since it ties so heavily into the multiplication framework.
Once this is merged, it would be nice to see how performance under julia 0.7/1.0 compares with the previous commit run under julia 0.6. Is there any overall change in speed? Related to #21.
|
Also note that there are a few warnings under CI: EDIT: sorry, I had forgotten that this warning is intentionally generated by this package; carry on. |
|
Regarding benchmarking, my (nonscientific) results indicate that the tests take ~70% longer to run on julia 0.7 (even with this pull request). I'm not even sure that the culprit is within Expokit.jl itself, but it would be nice to get to the bottom of this ... EDIT: perhaps related to JuliaLang/LinearAlgebra.jl#545 |
|
I agree that we should do a comparison under 0.6 and 0.7/1.0. The question is if we should merge this anyways to get rid of deprecation warnings etc and open a new issue to track the possible regression? |
|
I suggest that you merge and request to tag a new release, adding in the notes that this version drops support for v0.6; note that different versions can be benchmarked with julia> benchmarkpkg("Expokit", tagged-version-or-branch) |
Addresses JuliaLang/julia#33 : First I was trying to stay compatible to 0.6, but it was too much hassle. So with this PR support for versions < 0.7 is dropped.