Created attachment 1125
dustmite'ed test case
Another forward reference bug of a lurky nature, looks like it requires a
certian import chain to manifest. In fact there could be a collaterial bug(s)
at work here. See below.
Extract and compile files from attached source tree like this:
dmd std2/uni.d
output:
std2/internal/uni_tab.d(4): Error: struct std2.uni.RleBitSet!(uint).RleBitSet
no size yet for forward reference
dmd std2/regex.d
output:
std2/string.d(1): Error: module regex from file std2/regex.d conflicts with
another module regex from file std2/regex.d
dmd std2/string.d
output:
std2/functional.d(2): Error: module string from file std2/string.d conflicts
with another module string from file std2/string.d
Others show the same remarkably bogus diagnostic. It's on dmd 2.060 from git
HEAD. And it pretty much blocks my GSOC exploits unfortunately.
Comment #1 by dmitry.olsh — 2013-04-07T03:02:23Z
Looks like it was either fixed and diganostic improved.
Now it says something along the lines of:
std2/string.d(1): Error: module std2/regex should be imported as regex
And indeed these should all have proper module name. Once that is done it compiles just fine, so I'm closing it as WORKSFORME.