Proposal: Add variants of tails & inits not returning empty lists
Hi all, I just opened a Trac ticket (#1979), to suggest the addition of the inits1 and tails1 functions, which don't return empty lists unlike their counterpart. They're defined as such : inits1 = tail . inits tails1 = init . tails I've needed them a few times so I thought they would make a nice addition to the base package. And, I've got to say, the symmetry in this code is is nice :-). http://hackage.haskell.org/trac/ghc/ticket/1979 Cheers, Maxime
Hello Maxime, Friday, December 14, 2007, 7:17:38 PM, you wrote:
I just opened a Trac ticket (#1979), to suggest the addition of the
i have dozens of small funcs in my Utils.hs file. every experienced Haskeller has such set. are all these functions should be really added too? -- Best regards, Bulat mailto:Bulat.Ziganshin@gmail.com
On Dec 14, 2007 5:34 PM, Bulat Ziganshin
i have dozens of small funcs in my Utils.hs file. every experienced Haskeller has such set. are all these functions should be really added too?
Yes, if there is a lot of overlay in every experienced haskeller's set of small utility funcs :) Juanma
Bulat Ziganshin wrote:
Hello Maxime,
Friday, December 14, 2007, 7:17:38 PM, you wrote:
I just opened a Trac ticket (#1979), to suggest the addition of the
i have dozens of small funcs in my Utils.hs file. every experienced Haskeller has such set. are all these functions should be really added too?
I am entirely open to discussion about whether these two functions are useful enough to fit in the base libraries or not. I personally feel that they are, seeing that I had to use them several times in a relatively short period of time. I thus expect that many others have needed to write those functions as well. This may very well turn out to be a completely wrong :-). I guess one step towards is the answer would be knowing if you ever needed those functions? Cheers, Maxime
Hello Maxime, Friday, December 14, 2007, 10:53:26 PM, you wrote:
i have dozens of small funcs in my Utils.hs file. every experienced Haskeller has such set. are all these functions should be really added too?
I am entirely open to discussion about whether these two functions
well, are you have other common functions that are used in many projects? how many haskell code you've developed so much? are you plan to propose to add to stdlib every function that you used "a few times"? -- Best regards, Bulat mailto:Bulat.Ziganshin@gmail.com
Maxime Henrion wrote:
Bulat Ziganshin wrote:
Hello Maxime,
Friday, December 14, 2007, 7:17:38 PM, you wrote:
I just opened a Trac ticket (#1979), to suggest the addition of the i have dozens of small funcs in my Utils.hs file. every experienced Haskeller has such set. are all these functions should be really added too?
I am entirely open to discussion about whether these two functions are useful enough to fit in the base libraries or not. I personally feel that they are, seeing that I had to use them several times in a relatively short period of time. I thus expect that many others have needed to write those functions as well. This may very well turn out to be a completely wrong :-).
I guess one step towards is the answer would be knowing if you ever needed those functions?
For proposals like this, I think we need a significant number of people saying "yes please!", otherwise the default is for the code to go in without much discussion, and if we're not careful we end up adding things that are rarely useful. My data point: I've never used/needed inits1 or tails1, AFAIR, and I don't think "tail.inits" is too hard to write, so I vote against. Cheers, Simon
Hi
For proposals like this, I think we need a significant number of people saying "yes please!", otherwise the default is for the code to go in without much discussion, and if we're not careful we end up adding things that are rarely useful.
My data point: I've never used/needed inits1 or tails1, AFAIR, and I don't think "tail.inits" is too hard to write, so I vote against.
I vote against, unless new names can be found which give a more instant connection to the meaning. (I kind of said this in my original post, but should be a bit clearer I guess) Thanks Neil
On Mon, Jan 28, 2008 at 11:28:23AM +0000, Simon Marlow wrote:
For proposals like this, I think we need a significant number of people saying "yes please!", otherwise the default is for the code to go in without much discussion, and if we're not careful we end up adding things that are rarely useful.
Maybe some sort of quorum is needed. Like, some percentage of interested parties must make their feelings known before a final decision is made. Or the default should be 'no' on new additions unless a lot of people chime in saying yes. John -- John Meacham - ⑆repetae.net⑆john⑈
I've never needed them. So, no thanks.
Hi Maxime,
inits1 = tail . inits tails1 = init . tails
They look useful, and I have used ones like these before. However, the name is slightly misleading. foldr1 and foldl1 don't "not return the empty list", but both crash if given the empty list. These ones do something very similar and very different from the other ...1 functions, so giving them similar names might result in confusion. That said, I can't think of anything better. Thanks Neil
Neil Mitchell wrote:
Hi Maxime,
inits1 = tail . inits tails1 = init . tails
They look useful, and I have used ones like these before. However, the name is slightly misleading. foldr1 and foldl1 don't "not return the empty list", but both crash if given the empty list. These ones do something very similar and very different from the other ...1 functions, so giving them similar names might result in confusion. That said, I can't think of anything better.
Yeah, I agree those names aren't great and could lead to confusion. I have thought about calling them tails' and inits' but I think that would be worse, since the trailing single quote is used to indicate a more strict version most of the time and this has nothing to do with strictness. At least the tails1 and inits1 names carry the same "non-empty list" concept as foldl1 and foldr1, although in a different sense. Anyone has any better suggestion? Cheers, Maxime
participants (7)
-
Bulat Ziganshin -
John Meacham -
Juanma Barranquero -
Lennart Augustsson -
Maxime Henrion -
Neil Mitchell -
Simon Marlow