On Sunday, 2 November 2014 at 13:06:24 UTC, Dicebot wrote:
I had initial go at cleaning safety attributes :
https://github.com/Dicebot/phobos/tree/logger-fix-trust
It still needs one druntime PR merged
(https://github.com/D-Programming-Language/druntime/pull/1009)
to be done properly but should be ready for destruction.
David snipet results in this after all changes:
std/experimental/logger/core.d(1675): Error: safe function
'std.experimental.logger.core.Logger.logf!(11, "aaa.d",
"aaa.main", "void aaa.main() @safe", "aaa", Dangerous).logf'
cannot call system function
'std.format.formattedWrite!(MsgRange, char,
Dangerous).formattedWrite'
I don't recall off the top of my head some non-template innards
actually might require @safe, but apart from that, why not just
leave the job to template attribute inference entirely? If
somebody wants to log a type with a @system toString in non-@safe
code, why not just let them?
David
