On 12/17/2013 05:14 PM, Josh Stone wrote: > On 12/17/2013 01:33 PM, Bill Williams wrote: >> On 12/16/2013 07:12 PM, Josh Stone wrote: >>> On 12/16/2013 04:42 PM, Josh Stone wrote: >>>> Since the Objects and AObjects are not in public headers, I could just >>>> try making those into private unimplemented copy constructors, and see >>>> if anything breaks. >>> >>> This much appears to work fine, disabling copy and assignment for both. >>> (At least for the elf side; I don't have other systems to test.) >> >> Good, good. As I noted, the *design* is for ref-counted, and anyone >> trying to copy these would get a rude surprise (no symbols, for one!). > > I took a blind stab at Object-nt, at least, but Object-coff and > Object-xcoff seem to be obsolete, chaining to AObject constructors that > no longer exist. In coff it uses AObject(string,...) which was removed > in 2007 when MappedFile was introduced, commit 30930df02afa. In xcoff > it uses AObject(MappedFile*,MappedFile*,...) which was removed about a > year ago, commit f175cf6e5ff9. > > Nothing seems to reference Object-coff at all. Object-xcoff is only > pulled in for defined(os_aix) -- can I infer that neglect of AIX support > makes it effectively dead code? > > I'll ignore these two for now, since they're already broken. Pruning > unused platforms is a fun task for another day, if these are truly not > desired anymore. I love the smell of negative-LOC patches... :)
I've pushed the whole series. Someone with a Windows build environment should check that Object-nt works, at least, but I think it's ok. Don I now my flame apparel, in case anything breaks... Josh _______________________________________________ Dyninst-api mailing list Dyninst-api@cs.wisc.edu https://lists.cs.wisc.edu/mailman/listinfo/dyninst-api
