ojhunt wrote:

> Why is that test so long?

it's testing the surrounding issues: there are 1, 2, and 3 varargs arguments in 
the format string tests, and incrementing through the string changes the number 
of format specifiers. Then we test literals, const, and constexpr strings, and 
char buffers that are not null terminated.

Given the fix I want to make sure:
  * We correctly compute where the string starts
  * We correctly identify the length of the string
  * We correctly handle out of bounds strings
  * We correctly handle non-null terminated char buffers
  * We correctly handle the various different ways of computing a constant char 
buffer
  * We produce the same output with both the old and new interpreters
  
I don't like testing just the specific reported symptom, as that makes it easy 
to unnecessarily miss or regress trivially similar issues. This test covers all 
the new code paths and all the paths that could obviously lead to it - I'm not 
sure how I could make it smaller?

https://github.com/llvm/llvm-project/pull/173212
_______________________________________________
cfe-commits mailing list
[email protected]
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

Reply via email to