Page MenuHomePhabricator

SelectionDAGDumper: print datatypes behind node names
AbandonedPublic

Authored by MatzeB on Sep 2 2015, 10:54 AM.

Details

Reviewers
resistor
Summary

Before:
t2: i32,ch = load t0, t0, t1<LD4[FixedStack-1]> [ORD=1]
t7: i32 = Constant<0>
Now:
t2 = load:i32,ch t0, t0, t1 <LD4[FixedStack-1]> [ORD=1]
t7 = Constant:i32 <0>

Diff Detail

Repository
rL LLVM

Event Timeline

MatzeB updated this revision to Diff 33829.Sep 2 2015, 10:54 AM
MatzeB retitled this revision from to SelectionDAGDumper: print datatypes behind node names.
MatzeB updated this object.
MatzeB added a reviewer: resistor.
MatzeB set the repository for this revision to rL LLVM.
MatzeB added a subscriber: llvm-commits.

I'm not sure how much I care either way, but can't you just omit the '=' sign when printing the constants inline?

My thinking here was:

  • The datatype is just another attribute stored with the node, it is not an attribute of the magic value marker tXX
  • the output is consistent for inlined and non-inlined nodes.

I'm not sure how much I care either way, but can't you just omit the '=' sign when printing the constants inline?

Anyway I don't really care either for this (the inline printing is what I care about). So if this won't get a LGTM I'll change it to the suggested variant of just leaving out the equal sign when printing inline.

My thinking here was:

  • The datatype is just another attribute stored with the node, it is not an attribute of the magic value marker tXX
  • the output is consistent for inlined and non-inlined nodes.

I'm not sure how much I care either way, but can't you just omit the '=' sign when printing the constants inline?

Anyway I don't really care either for this (the inline printing is what I care about). So if this won't get a LGTM I'll change it to the suggested variant of just leaving out the equal sign when printing inline.

I think that I'd prefer that. I'm leaning toward saying that the node:type formatting actually makes things more difficult to read.

MatzeB abandoned this revision.Sep 21 2015, 11:32 AM

I adapted http://reviews.llvm.org/D12567 to not depend on this anymore.