> On 18 Apr 2018, at 11:19, Frank Heckenbach <f.heckenb...@fh-soft.de> wrote: > > Hans Åberg wrote: > >> Indeed. Others, may not want to patch their installations even it >> they can, as it may cause problems for other packages. Say if one >> writes package that uses a standard Bison installation, then it >> would be dangerous to patch it up with an experimental one. > > Not really dangerous, as my C++17 patch only adds new files. (My other > patches change existing files, but they are rather small, and just > bugfixes.)
It is your package, so you do with what you want. >>>> But if someone uses their of variants type, would that require C++11? >>> >>> Yes, the skeleton uses move semantics (i.e. rvalue references), so >>> C++11 is the minimum. (Not entirely sure if I accidentally used some >>> C++14 feature, as that's what I tested with besides C++17, but if >>> so, it may be rather easy to avoid.) >> >> Would it in principle possible to use ones own variants that does >> not require C++11 and put the C++11 code within preprocessor >> directives so that it works for earlier C++ versions? - I am not >> sure there would be any use for this, just wondering. > > Not (easily) with my patch, as it uses move semantics throughout. > For this use case, I think it would be easier to start with the > C++03 skeleton and change the variant implementation there. OK.