Our Windows expert, @bellenot, is on his deserved vacation at the moment. He’ll let you know the answer when he is back, in about two weeks from today! (All I can say is “try VC2015, that might not completely work but it will get you closest”.)
Do you know if there are a few other Windows users of cling/LLVM? I tried the Jupyter cling kernel last night, and it’s very Unix-bound. I’ve found some other things, and if Windows isn’t a concern at all, then it doesn’t seem worth is to investigate and submit the bugs discovered.
Would it help if I submitted bugs in JIRA for issues I find? I don’t really know enough about it all to dive into “Here’s the problem, here’s the fix.” But I can hopefully give some basic information.
Thanks for the offer! We have a pretty good idea what’s missing / currently broken, and we have a series of patches / contributions that we need to move into cling. So instead of Jira tickets it would probably be more relevant for us to hear from time to time (whenever you see my colleague Bertrand Bellenot merging patches from Mr “Frederich Munch”) whether things got better or worse on your side.
I’m a Java programmer by trade. I play around with C++ every now and then as a hobbyist.
Being of the opinion that jshell is the absolute best thing about JDK 9, I was so stoked when I googled for “c++ REPL” the other day, and discovered cling! Thanks for making such a neat tool available
I’m also using VS 2017 (Community edition). I just encountered the same error the OP is reporting.
Just as an aside — On my 7 year old Windows 7 personal laptop (8GB ram, 2.13 GHz i3 CPU) it wasn’t until the very end of like a 3+ hour build that the _MSC_VER test failed the build. Would there be any value in testing for the compiler version at the top of the build and failing earlier?
I’ll check what can be done, but the problem is not like for prerequisites, it only happens with newer version of Visual Studio, not yet defined in the source code (unrelated to the build system)
