[Bug default/28364] libwiretap fails self check with unspecified indirect sup-type changes

["dodji at redhat dot com" <sourceware-bugzilla@sourceware.org>]

https://sourceware.org/bugzilla/show_bug.cgi?id=28364

dodji at redhat dot com changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |ASSIGNED

--- Comment #1 from dodji at redhat dot com ---
I looked into this and got puzzled for a long while.

I think I know have a theory about what's going on.

I think something is going on in canonicalization of function types and it has
to do with the type canonicalization propagation optimization.  That one again.

To demonstrate what I mean, I put up a candidate patch that disables type
canonicalization for function types.  Function types are thus compared
structurally.  Thins are still fast because the sub-objects of function types
are canonicalized.  And this fixes the issue.  So I need to understand why
exactly type canonicalization on function types is failing on this binary.

I've put the candidate patch here
https://sourceware.org/git/?p=libabigail.git;a=shortlog;h=refs/heads/PR28364.

When I understand why exactly function types canonicalization is failing, we
should stop disabling function type canonicalization.

-- 
You are receiving this mail because:
You are on the CC list for the bug.