Document renumbering lines in macros #40955
Conversation
doc/src/manual/metaprogramming.md
Outdated
| @adds_more_lines_2 (macro with 1 method) | ||
| ``` | ||
|
|
||
| `__source__` will contain a line number pointing to the user's call site (see more information about `__source__` above). |
There was a problem hiding this comment.
I think it would be better to link to the "macro invocation" section where the __source__ arg is defined rather than just saying "above", since that isn't too descriptive.
Co-authored-by: Ian McInerney <Ian.S.McInerney@ieee.org>
|
Thanks for the suggestions! I was having trouble coming up with good terminology. |
|
Yea, choosing terminology is always a difficult thing to do. I noticed one last thing that needs to be cleaned up (the parenthesis was missing), and after that it looks good to me (but you need to wait for someone with commit rights to approve this as well). |
Co-authored-by: Ian McInerney <Ian.S.McInerney@ieee.org>
| This is for the following two reasons: | ||
|
|
||
| - If an error occurs in code introduced in a macro, you likely want stack-traces to point to the user's call site, rather than the macro's internals. | ||
| - If a user calls a macro, you likely want coverage to consider the user's call site as covered, rather than the macro itself. |
There was a problem hiding this comment.
These argument seem inaccurate to me. The source should already be be getting represented as the place where the macro is being inlined at (as a fake "caller"). Injecting it again into the macro body seems like it could lead to even stranger frames as it tries to guess why this macro is changing back and forth between inlined and local. If the source location is missing in the caller (where we splice the macro to), that seems like just a bug.
There's already Base.remove_linenums! and associated improvements in PR form: #31335 and Meta.replace_sourceloc! from #22623.
There was a problem hiding this comment.
If the source location is missing in the caller (where we splice the macro to), that seems like just a bug.
Ok, if I understand you right, a bug like this is likely what I'm running into
There was a problem hiding this comment.
Should I close this and open a new issue?
|
Replaced by #41043 |
So, this is something I've figured out on my own, so I'm not 100% sure if it's correct or not. If it's not correct, I'd greatly appreciate knowing what the correct way to handle this is.