[Bug translator/16464] New: Token should be a mandatory part of staptree constructors

["jistone at redhat dot com" <sourceware-bugzilla at sourceware dot org>]

https://sourceware.org/bugzilla/show_bug.cgi?id=16464

            Bug ID: 16464
           Summary: Token should be a mandatory part of staptree
                    constructors
           Product: systemtap
           Version: unspecified
            Status: NEW
          Severity: minor
          Priority: P2
         Component: translator
          Assignee: systemtap at sourceware dot org
          Reporter: jistone at redhat dot com

This is a cleanup task, with no particular known failure at the moment, but see
PR16166 for example where we've been bitten before.

Every staptree expression and statement should have an associated token, but
currently this is left to programmer diligence to assign it.  This really ought
to be set as part of the constructor.  This also requires trickling it down
from the constructors of all the inherited classes.

I believe it's also true that the chosen token of an expression or statement
will never change.  If so, we should also make the pointer itself const,
changing it to "const token* const".  We might even make it "const token&", to
emphasize its nonnull-ness, but that would increase the scope of refactoring.

-- 
You are receiving this mail because:
You are the assignee for the bug.