[PATCH] D71116: [test][tools] Add missing/Improve testing

[George Rimar via Phabricator via llvm-commits]

grimar added inline comments.


================
Comment at: llvm/test/tools/llvm-symbolizer/functions.s:17
 ## input address, and just prints it.
+# RUN: llvm-symbolizer 0 -f none --obj=%t.o | FileCheck %s --check-prefixes=LINKAGE,ERR
 # RUN: llvm-symbolizer 0 --functions none --obj=%t.o | FileCheck %s --check-prefixes=LINKAGE,ERR
----------------
jhenderson wrote:
> grimar wrote:
> > MaskRay wrote:
> > > Unrelated to this patch.
> > > 
> > > -f can be either --functions or --functions=, and the parsing accounts for the next option. This looks strange.
> > I am not sure I understood your concern.
> > It is just the same as `--functions none` but with the use of alias `-f`, isn't?
> > 
> > In another test we have a similar check:
> > 
> > ```
> > # Check --obj aliases --exe, -e
> > # RUN: llvm-symbolizer 0xa 0xb --exe=%t.o | FileCheck %s
> > # RUN: llvm-symbolizer 0xa 0xb -e %t.o | FileCheck %s
> > # RUN: llvm-symbolizer 0xa 0xb -e=%t.o | FileCheck %s
> > # RUN: llvm-symbolizer 0xa 0xb -e%t.o | FileCheck %s
> > ```
> Like @grimar, I'm not sure what it is you are concerned by?
> 
> Note that `--functions none` is NOT the same as `--functions=none`. The former is actually equivalent to `--functions=linkage none`. This block of tests demonstrates this for both --functions and -f (which should be identical in behaviour).
After re-reading the original comment today, I've got it I think.
Fangrui did not mean that this change is unrelated, but pointed to the difference of the behavior,
which is "--functions none is NOT the same as --functions=none". It is indeed looks strange
(I've never saw such things before I think).



Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D71116/new/

https://reviews.llvm.org/D71116