-
Notifications
You must be signed in to change notification settings - Fork 36
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
v0.6 precompile error: new UnionAll type breaks things #65
Comments
can be fixed with
but then fails in |
Subtypes() should accept UnionAll, not sure about everything else :( |
This should not work --- an array of types is not a valid type. I'll add an error for this. |
ColorTypes is basically a poster child for the kind of package that could benefit from the type revisions (e.g., Lines 177 to 257 in a0f9898
|
I have a work in progress for this package. |
The last status was that I hit JuliaLang/julia#18457 (comment). |
I also remember (I think) there being a number of places that could have benefited from triangular dispatch, which I think I worked around by expanding the number of arguments in several places. E.g., foo{C<:Colorant}(::Type{C}) = _foo(C, eltype(C))
# and now _foo can dispatch on both C and T where e.g. C = RGB{Float32}, T = Float32 |
Any progress on this? This breaks all dependent packages on 0.6 (including gadfly). |
This is also breaking |
this breaks basically everything :P julia> Pkg.dependents("Colors")
50-element Array{AbstractString,1}:
"Luxor"
"Images"
"ImageFiltering"
"VoronoiDelaunay"
"NamedColors"
"PlotUtils"
"VLFeat"
"Qwt"
"ProteinEnsembles"
"Cairo"
⋮
"QuartzImageIO"
"GraphPlot"
"ImageView"
"Brim"
"ColorBrewer"
"Winston"
"Caesar"
"FaceDatasets"
"Bio" |
I have a branch of julia/ColorTypes that passes here. However, I'll be traveling for the rest of the week, so I'm not gonna be able to clean this up. So here's the deal: I'll put up what I have in it's messy state (over the next hour or so), but with what I believe are all the tricky issues worked out and somebody can pick it up and finish it off. Otherwise, I'll get to it next week. I also have a PR for FixedPointNumbers, which this depends on. Same deal. |
Sounds good to me. I can't promise to get to it very quickly, but I might, and of course there seem to be other volunteers in this thread who could tackle it. Maybe easiest if you push as a branch rather than a gist? |
Pushed as kf/06 here and as a PR on FixedPointNumbers. This also requires my kf/newreflection branch of Julia itself. |
@Keno Any idea what will happen with JuliaLang/julia#20006? Simon is unfortunately right: ColorTypes.jl breaks an awful lot of testing on 0.6 |
Looks like it's actively being worked on. These type-system issues aren't simple. You could disable testing on nightly if the failures really bother you. |
It's through review. I'll merge it as soon as it's green on CI. Looks like the latest set of changes broke sth. |
That's awesome. Don't worry about the breakages, I can take a peek. Really appreciate your help on this. |
Sorry, what I meant was that my latest changes on that branch broke something in base, which I'll have to investigate. I did already rebase the new ColorTypes on top of the changes in base (but haven't pushed that yet). |
OK, just ping here when all is set (I am too swamped to follow what's happening with the type-system stuff). |
just to better track the progress here, this is the PR we are waiting for: |
It looks like the PR was merged, this means that the issue is solved in master? |
i think we're still waiting for
#69 to merge
On Tue, Feb 7, 2017, at 20:00, Júlio Hoffimann wrote:
It looks like the PR was merged, this means that the issue is solved
in master?
— You are receiving this because you commented. Reply to this email
directly, view it on GitHub[1], or mute the thread[2].
|
can be fixed with:
but then it fails again on
The text was updated successfully, but these errors were encountered: