Joseph Rushton Wakeling

> tl;dr: @WebDrake recently changed the rdmd to automatically add semicolons for the last statement without effect (#303) For the record (we've already discussed this in other PRs): no, I...

BTW, I would like to understand better the particular motivation behind this feature, other than that it seems cool. "95% of all uses" are mentioned above. What _are_ those uses?...

> Could you provide examples you would find surprising? If I ask for a piece of D code to be evaluated, I expect for that D code to be evaluated...

NO. This is fundamentally the wrong way to be thinking about this. `--eval` is a feature which has been part of `rdmd`, virtually unchanged (apart from bugfixes) for almost 10...

> I dont know why your assuming how Im thinking about this. I've defined my sense of "surprising" clearly: it's surprising if D code provided to `--eval` behaves differently from...

Folks, I'd like to ask that we put the focus on the questions I asked above: https://github.com/dlang/tools/pull/317#issuecomment-365777561

@wilzbach OK, but I still think it would be good to be clear about the motivation and _real_ use-cases if this is ever to be un-blocked. "It's a nice idea"...

> I'm currently thinking about limiting this to a smaller subset, e.g. insert a "writeln(...);" if there's no semi-colon at the end and the beginning of the eval code can...

@RazvanN7 I take it you meant @WebFreak001 there ... ? :-)

Sorry about earlier test failure: I forgot to tweak `win32.mak` ;-) Fixed now.