On Thursday, 16 March 2017 at 22:42:21 UTC, Jonathan M Davis
wrote:
On Thursday, March 16, 2017 21:50:18 Sebastiaan Koppe via The
problem is that that then catches when an enum is passed to
isSomeString, not when the template constraint or static if
that isSomeString is being used in should have checked for
enums and didn't. And checking isSomeString with an enum is
perfectly valid. It's just that we'd like it to be false,
whereas right now it's true. Code could quite legitimately be
doing something like
auto foo(T)(T str)
if(isSomeString!T)
{
static if(is(T == enum))
return foo!(StringTypeOf!T)(str);
else
{
...
}
}
Ahh yes, silly me. It's not so much about isSomeString but how
people use it in conjunction with extra constraints. Quite messy.
So for my proposed static deprecated to work it would only have
to show a message when it would have changed the outcome of the
template constraint where it is used in. Which still gives false
positives due to overloads.
So the only way to change a template trait is to either just do
it - and risk breakage - or to create a new one under another
name?
So, probably, all we could do is figure out how often
isSomeString is used in the D code that's publicly up on places
like github or bitbucket and not much about how it's used. I
don't know for sure though.
- Jonathan M Davis
Yeah, you are right. You'll would need to instantiate the
templates to resolve the types and as far as I know dcd/scanner
don't.
