On Mon, 24 Feb 2014 17:22:32 -0800, Meta <jared...@gmail.com> wrote:
On Tuesday, 25 February 2014 at 00:28:26 UTC, Adam Wilson wrote:
[SNIP]
You're throwing what I said way out of proportion. I was replying to the
statement:
"No criticism should stop this module being accepted, as we do not
have any other lexer in the runtime anyway. Therefore I suggest
we accept std.lexer until a better solution comes up."
I don't agree with this. Obviously std.lexer is well-written and has
been through a few rounds of iteration, but that doesn't mean we should
accept it without due diligence to ensure that we won't be regretting
some overlooked, poorly-designed or badly-named piece of functionality
down the road. "Good enough because we don't yet have anything better"
is a bad idea. It seems to me that what Brian has written is much better
than "good enough", but I don't think that it should be accepted into
Phobos *solely* because we don't have anything else. If the community
decides that it is a worthwhile addition, then great, but that must not
happen *until* it has passed rigorous review, just like every other
recent Phobos module.
Fair enough. I guess I am just still touchy after the way std.signals was
shot down. There weren't great technical arguments for shooting it down
and so I feel that a good piece of code that would've been immediately
useful and accepted by the community was rejected over some pretty silly
fears.
Note that I as badly as I want std.lexer to be included I want it to pass
a rigorous review. This review (and, is passing, subsequent inclusion) has
opened up an opportunity to start using D at work that I did not expect
and so I am kind of excited about it.
--
Adam Wilson
GitHub/IRC: LightBender
Aurora Project Coordinator