inlined code line numbers need work #13725
Labels
bug
Indicates an unexpected problem or unintended behavior
error handling
Handling of exceptions by Julia or the user
regression
Regression in behavior compared to a previous version
Comments
JeffBezanson
added
error handling
|
Ref my other example in #13553 (comment) |
|
We would need to handle line nodes spliced in by macros too, unless these enter/pop things are labeled. Ref: #1334 (comment) |
|
I think we might have to put back the line number filter in inlining. I'm seeing really extreme regressions in line number info; for example looking at #13350 I get the following profile full of random locations: Putting back the line number filter, I get the following accurate and useful output: |
JeffBezanson
added
regression
|
Ok, I think this particular issue is not present in 0.4 thankfully. |
|
though it likely would if #13491 were backported? |
|
Correct, especially commit a33cd07 |
JeffBezanson
added a commit
that referenced
this issue
Oct 26, 2015
#13725 Single expression functions tend to get inlined into the middle of an expression, so their line numbers should not be inserted into the outer function (the number applies to only part of an expression, not the whole statement). This also significantly reduces sysimg size.
bjarthur
pushed a commit
to bjarthur/julia
that referenced
this issue
Oct 27, 2015
JuliaLang#13725 Single expression functions tend to get inlined into the middle of an expression, so their line numbers should not be inserted into the outer function (the number applies to only part of an expression, not the whole statement). This also significantly reduces sysimg size.
|
using Base.Test
for i = 1:4
@test 1 == 2
end
_
_ _ _(_)_ | A fresh approach to technical computing
(_) | (_) (_) | Documentation: http://docs.julialang.org
_ _ _| |_ __ _ | Type "help()" for help.
| | | | | | |/ _` | |
| | |_| | | | (_| | | Version 0.3.12-pre+7 (2015-10-20 12:08 UTC)
_/ |\__'_|_|_|\__'_| | Commit fa31586* (10 days old release-0.3)
|__/ | x86_64-apple-darwin15.0.0
julia> include("/Users/andreasnoack/Desktop/testfile.jl")
ERROR: test failed: 1 == 2
in error at error.jl:21
in default_handler at test.jl:19
in do_test at test.jl:39
in anonymous at no file:5
in include at ./boot.jl:245
in include_from_node1 at ./loading.jl:128
while loading /Users/andreasnoack/Desktop/testfile.jl, in expression starting on line 3
_
_ _ _(_)_ | A fresh approach to technical computing
(_) | (_) (_) | Documentation: http://docs.julialang.org
_ _ _| |_ __ _ | Type "?help" for help.
| | | | | | |/ _` | |
| | |_| | | | (_| | | Version 0.4.1-pre+21 (2015-10-29 07:02 UTC)
_/ |\__'_|_|_|\__'_| | Commit be50c8d* (1 day old release-0.4)
|__/ | x86_64-apple-darwin15.0.0
julia> include("/Users/andreasnoack/Desktop/testfile.jl")
ERROR: LoadError: test failed: 1 == 2
in expression: 1 == 2
in error at ./error.jl:21
in default_handler at test.jl:30
in do_test at test.jl:53
[inlined code] from /Users/andreasnoack/Desktop/testfile.jl:5
in anonymous at no file:0
in include at ./boot.jl:261
in include_from_node1 at ./loading.jl:304
while loading /Users/andreasnoack/Desktop/testfile.jl, in expression starting on line 3Update: I still think there is an issue, but the example was bad because the right line number is in the output here. I tried to make a small reproducible example, but it didn't capture the issue. I'll try to make a new example. |
|
fixed by #14949 |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Example (encountered while working on #13547):
The line
abstractarray.jl:301is correct, but the other 3 line numbers are completely wrong. I believe the fix is for the inliner to insert some kind of location "pop" operation. For example we haveWe see several line nodes piled up, without knowing their nesting structure.
The text was updated successfully, but these errors were encountered: