Proposal: Enum.reject for Enum.compact

[Bruce Tate]

Every couple of months, we get a request for Enum.compact. So far, the main argument against it is that Enum.reject list, &is_nil/1 is sufficient. 

When you compare the difference in a pipe, you can see a significant difference in readability. 

It really is a pervasive use case. 

list

|> nil_or_result

|> compact

versus

list 

|> nil_or_result

|> reject(&isnil/1)

One possible solution is for Enum.reject to take a default function as an argument. I suggest a default of  fn(x) -> x end

That would make the above example: 

list

|> nil_or_result

|> reject

which reads just fine, I think. There is ample precedent for meaningful defaults in enum. is_nil/1 and &(&1) are meaningful defaults. 

Thoughts? 

-bt

--

Bruce Tate
President, RapidRed, LLC
Phone: 512.772.4312
Fax: 512 857-0415

Author of Seven Languages in Seven Weeks, Deploying Rails Applications, From Java to Ruby, Rails: Up and Running, Beyond Java, 6 others.

[Tallak Tveide]

I assume you mean the default function should be

'''
fn x -> !x end
'''

To reject any falsey value. It does seem reasonable, but I dont think it is nearly as readable as 'compact', so I vote nay to this one...

In fact, isn't this a better option? (Or equally good)

'''
def filter(enum, fun \\ fn x -> x end)
'''

Actually I have also thought that a 'unity' or 'itself' function should be part of stdlib, to make this possible as:

'''
list
|> Enum.filter(&itself)
'''

[Louis Pop]

Hiya

I'd like an identity function in Kernel, that's for sure.

I think an identity function being the default argument to filter would be quite nice. I find reject a bit strange, I don't set why it exists when we have filter.

Cheers,
Louis

--
You received this message because you are subscribed to the Google Groups "elixir-lang-core" group.
To unsubscribe from this group and stop receiving emails from it, send an email to elixir-lang-co...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/elixir-lang-core/06bc7738-3ed9-4a21-9fce-b7eb3fecd600%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[OvermindDL1]

`identity : 'a -> 'a` is the most common function declaration like that in almost every functional language, would be nice to have in kernel.

[Drew Olson]

I've also often wished for an identity/1 in Kernel, especially because you cannot easily represent it with the short function syntax.

To unsubscribe from this group and stop receiving emails from it, send an email to elixir-lang-core+unsubscribe@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/elixir-lang-core/07571247-7dba-4a21-948d-89f7ac335f92%40googlegroups.com.

[Drew Olson]

While we're wishing for things, I've also wanted constant/1 a few times:

constant(3).()

# => 3

Especially for things like creation of Agents.

[Drew Olson]

Ignore my comment about not being about to use the shorter function syntax. I was confused about the precedence of &.

On Thu, Nov 17, 2016 at 1:08 PM, Drew Olson <dr...@drewolson.org> wrote:

[OvermindDL1]

Yeah you could do `&(&1)` or so for an inline identity, though definitely not as clear.  :-)

Constant is not as succinct though:  `&(fn->&1 end)`

And still not clear.  ^.^

[Bruce Tate]

The identify function is effectively already the default for Enum.all? and any?, and a few others as well. 

Why not make it the default for Enum.reject too?

-bt

To unsubscribe from this group and stop receiving emails from it, send an email to elixir-lang-core+unsubscribe@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/elixir-lang-core/8c4a990c-77be-4d00-857b-9a26c59715f9%40googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

--

[Peter Hamilton]

Bruce: As pointed out previously, It's a bad default for Enum.reject. It would leave you with just the nil values. It might be an ok default for Enum.filter though.

To view this discussion on the web visit https://groups.google.com/d/msgid/elixir-lang-core/8c4a990c-77be-4d00-857b-9a26c59715f9%40googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

--

Bruce Tate
President, RapidRed, LLC
Phone: 512.772.4312
Fax: 512 857-0415

Author of Seven Languages in Seven Weeks, Deploying Rails Applications, From Java to Ruby, Rails: Up and Running, Beyond Java, 6 others.

--

You received this message because you are subscribed to the Google Groups "elixir-lang-core" group.
To unsubscribe from this group and stop receiving emails from it, send an email to elixir-lang-co...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/elixir-lang-core/CAMp4_igw0oQsX3Q0i8KZ0mZqBMsdsxdzMGtSM-r4eGa7QtS5Yg%40mail.gmail.com.

[Chris Keathley]

It seems like an identity function would be a better default for both Enum.filter and Enum.reject. Then instead of using Enum.reject(&is_nil/1) you would use Enum.filter. I'm not sure if defaulting to identity could return surprising results. But it seems like it might be OK since thats what Enum.any? and Enum.all? already do.

[Viza-]

Yes, would be really handy to have these features for .filter and .reject


четверг, 17 ноября 2016 г., 18:35:59 UTC+3 пользователь Bruce Tate написал: