Added ability to convert from Arrays of Nullables #99
Conversation
|
Hmm... I only checked Julia 0.4. Did I cause the nightly build test to break, or was it broken already? |
|
|
||
| #----- Conversion from Array of Nullables #1 -----------------------------------# | ||
| # Take an Array of Nullable objects and convert it into a NullableArray | ||
| function Base.convert{S,T,N}(::Type{NullableArray{S,N}},in::Array{Nullable{T},N}) |
There was a problem hiding this comment.
Please add spaces after commas as in the rest of the file.
There was a problem hiding this comment.
Ah, and you don't need the T here and below.
There was a problem hiding this comment.
Thanks for the code review, @nalimilan. How will it work without T? For example, we don't have Array{Nullable{Int,N}} <: Array{Nullable}, and I didn't want this function to capture Arrays that were filled with something non-Nullable.
There was a problem hiding this comment.
I meant in::Array{Nullable,N}, but now I see that you need to keep T in {S,T,N} anyway, so you can as well keep it.
|
Thanks. The Travis failure doesn't look related, but I guess it should be fixed anyway. |
| out | ||
| end | ||
|
|
||
| #----- Conversion from Array of Nullables #1 -----------------------------------# |
There was a problem hiding this comment.
Duplicate of the method above?
|
I would like to follow best practices and not merge in the presence of even seemingly unrelated failures. We need to do a pretty thorough review of how Base Julia is impacting the present package. If people want to help me sleuth through this, I'd welcome it. I have finals next week and won't be able to sit down and seriously examine these issues till after then. EDIT: I will mention I appreciate this PR and that I don't like holding up merging it on account of what are probably unrelated failures. |
|
Actually, @nalimilan @johnmyleswhite do you have any opinions about merging this once it's ready despite the state of this package on v0.5? I can put something in the README to emphasize that we don't claim to support 0.5 yet. |
|
As long as the tests pass on 0.4, it should be fine (unless the tests pass on 0.5 without this PR of course). Though there are still a few unadressed points. |
|
Yes, there are. But once those are addressed, I think I'll be comfortable merging. The breakage points are not new to this PR. Thanks for weighing in, @nalimilan . |
|
Agree with Milan that you should focus on making sure tests pass on 0.4. Targeting 0.5 gives you a moving target. |
|
Indeed. Okay, PRs need pass tests only on 0.4. Thank you for your thoughts, John. |
|
@andyferris Would you mind rebasing your PR? We were quite close from merging when the discussion stopped. |
|
Sorry, yes I have been distracted by many things (moving house, etc) and this one went on the back-burner. I'll try have a look tomorrow! |
andyferris
force-pushed
the
pull-request/84efecbd
branch
from
f00c994
to
406851b
Compare
|
Changes Unknown when pulling 406851b on andyferris:pull-request/84efecbd into * on JuliaStats:master*. |
andyferris
force-pushed
the
pull-request/84efecbd
branch
from
406851b
to
af609b1
Compare
|
OK, rebased and removed an duplicated function. Tests pending, but could be good to go? |
|
Changes Unknown when pulling af609b1 on andyferris:pull-request/84efecbd into * on JuliaStats:master*. |
|
Thanks! Looks good to me, if @davidagold wants to merge. Though you should probably squash the last commit. |
|
Yes, would you please squash those into one commit? I'd greatly appreciate that =) Otherwise, this looks good to go. |
|
Thanks, @andyferris! I've squashed them directly using the new GitHub support for that. |
These methods add the ability to create a
NullableArrayfrom data that is in anArray{Nullable{T},N}. Tests added.See #98