On Tuesday, 18 March 2014 at 07:45:03 UTC, Jacob Carlborg wrote:
On Monday, 17 March 2014 at 20:31:36 UTC, Andrei Alexandrescu
wrote:
I had the same concerns about it being front and center in
std. Now that it's internal that issue disappears - we can use
it inside Phobos for a while and change it without disrupting
users. In many ways putting it up for review after all makes
things better for everybody.
I don't think that issue disappears. Walter has done this
before and now he did it again. Creating a pull request for a
new module without asking for a formal review. It's less of a
problem that it was merged as an internal module. The big issue
trying to get everyone to follow the protocols we have for
adding new modules.
In his defense, he *is* trying to keep our modules clean, which
implies creating a new module. But let's put this into
perspective:
Said module is *1* type only. It's not like it was a new 10_000
line module or anything: It's 100 lines + unittest.
If it was me, I would have just put the damn thing as a new type
in "std.array", submitted it as a pull, with request for review
nor mention in the boards, and then be done with it. And *then*,
if we weren't happy about it, I would have marked it package too.
It's not the first time we do this too. There's a fair bit in
things in phobos that are "fairly useful, but don't justify
public exposure". This is just one more of them. It just happens
to be in a module.
I'm not entirely happy about the way it went down, but I think we
should cut him a little slack in that regards.
That, and I think there was bad communication. On both sides.
