Skip to content
This repository was archived by the owner on May 4, 2019. It is now read-only.

Add missing LineNumberNode to @swappable call (fixes #257)#259

Merged
ararslan merged 3 commits intoJuliaStats:masterfrom
getzdan:pull-request/7d602fa4
May 30, 2017
Merged

Add missing LineNumberNode to @swappable call (fixes #257)#259
ararslan merged 3 commits intoJuliaStats:masterfrom
getzdan:pull-request/7d602fa4

Conversation

@getzdan
Copy link
Copy Markdown
Contributor

@getzdan getzdan commented May 29, 2017

There has been some change to Julia in the macrocall AST, adding the LineNumberNode.
Unless there is another change to Julia, some packages using :macrocall expression generation will break.

@getzdan getzdan mentioned this pull request May 29, 2017
Closed
@nalimilan
Copy link
Copy Markdown
Member

nalimilan commented May 29, 2017

Thanks. Indeed this fixes loading the package, but I wonder whether the test failure may be related: the path between #= and =# is suspicious.

Booleans: Error During Test
  Expression evaluated to non-Boolean
  Expression: any(#= C:\Users\appveyor\.julia\v0.7\DataArrays\test\booleans.jl:15 =# @data([NA, 1, 2]) .== 1) == true
       Value: NA

@getzdan
Copy link
Copy Markdown
Contributor Author

getzdan commented May 29, 2017

Don't think the test error is related. Actually, it seems to me the test should Fail (but not error), as 1 == NA should be NA. But this discussion is for the other issue.

nalimilan
nalimilan reviewed May 29, 2017
View reviewed changes
Comment thread src/operators.jl Outdated
# For /, Array/Number is valid but not Number/Array
# All other operators should be swappable
map!(x->Expr(:macrocall, Symbol("@swappable"), LineNumberNode(@__LINE__), x, scalarfunc), fns, fns)
if VERSION < v"0.7.0-"
Copy link
Copy Markdown
Member

@nalimilan nalimilan May 29, 2017

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

It would be great if you could identify the exact commit which broke this (use contrib/commit-name.sh).

Copy link
Copy Markdown
Contributor Author

@getzdan getzdan May 29, 2017

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

My guess is the exact commit is: 0.7.0-DEV.328 . Specifically, JuliaLang/julia@fcdf437

Copy link
Copy Markdown
Contributor

@tkelman tkelman May 30, 2017 •
edited
Loading

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

You should use the merge commit JuliaLang/julia@9e3318c. Its master parent commit, JuliaLang/julia@8b2cac4, is 0.7.0-DEV.353 but does not have this change.

Copy link
Copy Markdown
Member

@nalimilan nalimilan May 30, 2017

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Damn, everybody gets this wrong at least once. Do you think the script could do this automatically?

Copy link
Copy Markdown
Contributor

@tkelman tkelman May 30, 2017

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

we could maybe have it do some git operations to try to figure out whether the commit it runs on is a merge (or squash merge) commit or something from a PR branch and give a warning, but I'm not sure what to check for

Copy link
Copy Markdown
Contributor

@tkelman tkelman May 30, 2017

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

If github would fix "rebase and merge" to include a reference to the PR number, we could switch to that for multi-commit PR's and disable the normal merge commit option. Single-commit PR's should almost always be squash merged since there's no reason not to, so it's only multi-commit PR's where there's a risk of VERSION getting things wrong.

Copy link
Copy Markdown
Contributor

@tkelman tkelman Jun 16, 2017

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

did this get fixed?

Copy link
Copy Markdown
Member

@nalimilan nalimilan Jun 16, 2017

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

No, but see #264.

@getzdan
Copy link
Copy Markdown
Contributor Author

getzdan commented May 29, 2017

You are right .== is acting weird:

julia> (@data [NA, 1, 2]) .== 1
0-dimensional DataArrays.DataArray{Bool,0}:
NA

@nalimilan
Copy link
Copy Markdown
Member

nalimilan commented May 29, 2017

I guess we can still merge this (after updating the version check) and continue debugging, as it's always easier to debug a package that can be loaded.

@ararslan
Copy link
Copy Markdown
Member

ararslan commented May 30, 2017

Thanks for figuring this out!

@ararslan ararslan merged commit 1521237 into JuliaStats:master May 30, 2017
This was referenced May 30, 2017
Merged
Merged
@nalimilan
Copy link
Copy Markdown
Member

nalimilan commented May 30, 2017

The failure actually comes from a change in Base, see JuliaLang/julia#22019 and JuliaLang/julia#22130.

@getzdan getzdan deleted the pull-request/7d602fa4 branch May 30, 2017 09:40
@nalimilan nalimilan mentioned this pull request Jun 16, 2017
Merged
Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.

Reviewers

2 more reviewers

@nalimilan nalimilan nalimilan left review comments

@tkelman tkelman tkelman left review comments

Reviewers whose approvals may not affect merge requirements

Assignees

No one assigned

Labels

None yet

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

4 participants

@getzdan @nalimilan @ararslan @tkelman