UndefVarError line number wrong [0.4 only]

[physicsd00d]

I'm using v0.4.0.
If I put the following code in a file (scopeError.jl):

for ix in 1:3
    if ix == 2
        y = 3
        print("y = $y\n")
    else
        print("$ix else\n")
        # print("y = $y\n")
    end
end

then run it, I get the expected output:

$ julia scopeError.jl 
1 else
y = 3
3 else

If I uncomment line 7 with the print("y = $y\n") and run it again, I get an error.

$ julia scopeError.jl 
1 else
ERROR: LoadError: UndefVarError: y not defined
 [inlined code] from /Volumes/Storage/Research/Help/scopeError.jl:4
 in anonymous at no file:3
 in include at /Applications/Julia-0.4.0.app/Contents/Resources/julia/lib/julia/sys.dylib
 in include_from_node1 at /Applications/Julia-0.4.0.app/Contents/Resources/julia/lib/julia/sys.dylib
 in process_options at /Applications/Julia-0.4.0.app/Contents/Resources/julia/lib/julia/sys.dylib
 in _start at /Applications/Julia-0.4.0.app/Contents/Resources/julia/lib/julia/sys.dylib
while loading /Volumes/Storage/Research/Help/scopeError.jl, in expression starting on line 1

You'll notice that it's telling me the error is on line 4, inside the if-clause immediately after y's initialization, and not on line 7 inside the else-clause where y is actualy undefined (until ix == 2). At the least, the error message should point to line 7 I think. For the equivalent code, Python puts the error at line 7.

What REALLY confuses me is that the documentation specifically states that if-blocks "do not introduce new scope blocks." In that case, it seems like y should be scoped inside the entire for-loop after it gets initialized at ix==2. Perhaps I'm too used to Python and that's not how it actually works in julia, but I would have expected the following code to run without any errors:

for ix in 1:3
    if ix == 1     # Notice that I changed this from 2 to 1, so y should be initialized for every run through the loop
        y = 3
        print("y = $y\n")
    else
        print("$ix else\n")
        print("y = $y\n")
    end
end

Yet it returns:

y = 3
2 else
ERROR: LoadError: UndefVarError: y not defined
 [inlined code] from /Volumes/Storage/Research/Help/scopeError.jl:4
 in anonymous at no file:3
 in include at /Applications/Julia-0.4.0.app/Contents/Resources/julia/lib/julia/sys.dylib
 in include_from_node1 at /Applications/Julia-0.4.0.app/Contents/Resources/julia/lib/julia/sys.dylib
 in process_options at /Applications/Julia-0.4.0.app/Contents/Resources/julia/lib/julia/sys.dylib
 in _start at /Applications/Julia-0.4.0.app/Contents/Resources/julia/lib/julia/sys.dylib
while loading /Volumes/Storage/Research/Help/scopeError.jl, in expression starting on line 1

The equivalent code works without error in Python. Notice that the above error is again pointing to the wrong place.

[yuyichao]

if does not introduce new scope but for does. The y in each loop are independent with each other.

I don't see the line number issue on 0.5.0-1108 with LLVM 3.7.

[vtjnash]

see #13491 for the line number fix

[tkelman]

Same opinion here as #13818 (comment)

This might be a wontfix unless someone can come up with a less-breaking-than-#13491 fix specifically for release-0.4.

[vtjnash]

the line that fixes the bug is https://github.com/JuliaLang/julia/pull/13491/files#diff-6d4d21428a67320600faf5a1a9f3a16aR5911, the rest of the PR was removing old workarounds that were no longer necessary

[tkelman]

should be fixed by #15385