groovy-users mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From OC <...@ocs.cz>
Subject Re: .with() variant that returns the original object
Date Tue, 08 Nov 2016 15:15:12 GMT
Hi there,

as always I can be wrong, but I do not think it is possible to find a good name which (a)
expresses the behaviour well (b) along with its very close similarity to 'with', which is
a problem of 'tap's etc can be found.

Therefore myself, I would advocate for

(a) **renaming current 'with'** (keeping, of course, the original name as an equivalent for
backward compatibility; just by convention discourage its use in new code)
(b) and adding an appropriate variant of the name for the new functionality.

Myself, I would probably like

- 'doWith' for the new one -- considering returning 'this' a reasonable default, which it
is
- 'valueWith' as a modern equivalent of 'with' for the old functionality

or perhaps

- 'tap' for the new one (for all the reasons others expressed in favour of the name)
- 'valueTapping' as a modern equivalent of 'with' for the old functionality

or something like that.

All the best,
OC


On 8. 11. 2016, at 15:34, Paul King <paulk@asert.com.au> wrote:

> Hi everyone,
> 
> We are hoping to release 2.5 not too far down the track. We are
> working on a revamped release process that is going to dramatically
> improve our ability to release early/release often but also comply
> with some additional Apache requirements that we follow these days.
> But more on that another time.
> 
> One outstanding feature request targeted for potential inclusion in
> 2.5 is an alternative to .with{} that automatically returns the
> original object after executing the closure - recall that .with{}
> returns the last evaluated expression. The proposal is here:
> 
> https://github.com/apache/groovy/pull/174/files
> 
> We can't use the old name since that would break backward
> compatibility and is of utility in its current form in any case, so we
> are looking for a new name for the proposed method. If you look at the
> PR you will see it has been called 'tap' and 'doto' and numerous other
> names have been suggested. We regard naming as very important and
> normally we'd have very strong views about suitable names based on
> functionality and similar method names within the Groovy codebase. But
> in this instance, an obvious name hasn't popped out at us, so we are
> requesting feedback from the community about what names make most
> sense to you.
> 
> Firstly, here is what the method does. I'll use 'tap' in these
> examples since that is what the PR currently uses but we can easily
> change that based on feedback.
> 
> myObject.tap {
>    // some code
> }
> 
> is equivalent to:
> 
> myObject.with {
>    // some code
>    return this
> }
> 
> Returning the 'self' object lends itself to various kinds of chaining, e.g.
> 
> assert [:].tap {
>    a = 1
> }.tap {
>    b = 2
> } == [a:1, b:2]
> 
> Or this one (adapted from a blog post[1] - and assuming you have a
> config.properties file containing answer=42 as one of the properties):
> 
> assert new Properties().tap {
>    new FileInputStream('config.properties').withCloseable {
>        load(it)
>    }
> }.answer == '42'
> 
> Here are some of the names that have been suggested with some commentary:
> 
> doto    Used by Clojure. Not camel case as per normal convention
> (though we have upto and downto which also break that convention) and
> it isn't immediately obvious which is which between with and doto just
> from the names
> 
> tap    Comes from Ruby and a previous Groovy extension outside core
> exists; meant to conjure up the idea of tapping into an object
> 
> autoWith    Same as with but automatically returns self
> 
> doWith   Again, hard to distinguish between doWith and with from the
> names themselves
> 
> tee    An alternative name for tap
> 
> auto    A shortened version of 'autoWith'
> 
> apply    same as Kotlin which has copied Groovy's with but suffers
> from the downside that apply is already heavily overleaded in other
> contexts, e.g. functional programming
> 
> withThis    Distinction with normal with a bit subtle perhaps
> 
> asThis    Ditto
> 
> within    Ditto
> 
> I'll also point out the 'identity' is currently an alias for 'with',
> in case that provides any additional inspiration.
> 
> Perhaps you dis/like one of the above or have some other suggestions.
> Let us know.
> 
> 
> Cheers, Paul.
> 
> 
> [1] http://beust.com/weblog/2015/10/30/exploring-the-kotlin-standard-library/


Mime
View raw message