Allow an attr_ variant for query-methods that end with a question mark '?' character, such as: def foo? returning @foo
Hi nobu :)
Also hi matz if matz reads this, and of course the rest of the core
team and everyone else.
Today on IRC, this mini-discussion happened (I show a snippet):
<apeiros> I really miss attr_query or whatever you want to name it <apeiros> which would generate a ? method too <jhass> apeiros: crystal has :P getter? <apeiros> nice
Ok, so the language crystal has something ruby does not have.
We can't let those newcomers get away with making ruby look old
now can we!
I use ruby not crystal but I very often use methods that end
with a '?' query mark in ruby. It helps me in simple if clauses
if hash.has_key? if hash.key? if cat.is_hungry?
(In the latter, it might be a cat of class
Cat instance, with
an instance variable called
@is_hungry, and when the cat is
fed with food, it is not hungry logically.)
We can generate these
attr_* right now as is
already, such as:
attr_reader :foo def foo; @foo; end attr_writer :foo def foo=(i); @foo = i; end attr_accessor :foo ^^^ Combines the above two methods into one.
But we have no way to designate methods that end via '?'.
I do not know which API call would be nice. apeiros on
I am fine with that. (The name is secondary for me, I
would like to have this feature available - what name
it would then have is not the main issue for me.)
apeiros then also suggested this syntax:
attr_* that would end with a
? token, would be a
attr_reader and also a variant of the
above that has a '?' token, so for example:
Would create both a method
People who do not need this, can continue to use:
So perhaps this suggestion is even better than
a new method (such as through
(I also have added one more line from apeiros,
not sure if I understood it, but I think the
above explanation should suffice - here is the
other suggestion he did:)
apeiros> e.g. attr_reader :foo? -> foo? // attr_accessor :foo? -> foo= + foo? // all with @foo of course. and foo? returning true/false.
Ok, that's it.
Thanks for reading!
Updated by spatulasnout (B Kelly) about 5 years ago
Yukihiro Matsumoto wrote:
Is there any real-world use-case for the proposal?
I've wanted this feature for years. :)
Here's an example from code I'm working with today:
def startup_complete? @startup_complete end
I would have preferred:
def foo? @foo end
is just fine.
Perhaps fine, but clunky. :)
In other words, why do we ever use
attr_reader (etc.) at all?
attr_.* because it's preferable to writing a bunch of
The reasons are identical for the '?' variant.
Updated by duerst (Martin Dürst) about 5 years ago
Daniel Berger wrote:
Abandoned all hope:
I think that this shows that there are several people who really like to use a foo? getter with a foo= setter. What it doesn't show that there are enough such people that it would balance out the confusion for those who think that a foo= setter goes together with a foo getter, at least by default.
With the current state, the main case is covered, nobody gets confused, and those who want a foo?/foo= pair can easily get it either as Matz showed above (easily reduced to one line) or by redefining the attr_... methods; the later is also very easy as it is usually about the first example given when explaining metaprogramming.