Talk:Search a list of records

The BugMeNot junk "task" assault (from behind a login control evasion service) continues :-)

Someone's desk job to establish beachhead positions ?

Maybe, but the credentials should not be as important as the content. Deleting due to the use of BugMeNot isn't a very good reason for deletion, although in these cases the content isn't so good either (but may be improvable; it is wiki so anyone can and should make such improvement please). --Zzo38 (talk) 19:53, 9 October 2015 (UTC)

draft task

Please see Add a Task and if you are more generally interested, check out the "Ways to Contribute" column on the home page. --Rdm (talk) 04:49, 10 October 2015 (UTC)

perhaps 2/3 slight expansions of the task description and requirements ?

The references to:

1. homogenous vs heterogenous lists, and
2. looking for elements of two different types

both allude to the implicit presence of specific matching/equality functions.

To get a good Rosetta task, in which comparison across languages makes the hidden convergences and divergences a bit more visible, and intelligible, perhaps ask submitters to show:

• The behaviour of the implicit equality function used by any built-in search function. (Case sensitive ? Int values = real values ? Expressions evaluated before the match ?)
• Show a search function which takes a coder-supplied equality function as an argument. (Finding by a particular property, given a list/array of objects (or otherwise compound elements), for example)
• A search which returns an index, and a search which return a value.

(As a footnote on apparent divergence and hidden convergence, Haskell, at first sight, doesn't use heterogenous lists, but it does use homogenous lists of compound data types. JavaScript appears to have heterogenous list of elements, which might include a mixture of reals, strings, functions, and objects, but in fact these are all JS objects, some atomic on the surface but all, in fact, compound below the water-line, each with a type/Constructor as well as various properties/values/sub-expressions. Searching for an item will always involve a function which derives a type-dependent value and applies a particular type of equality check, often involving some kind of reduction, normalisation or coercion.

Search also varies with evaluation strategy – if one element of a list is a function call, in some languages that call will be eagerly reduced/evaluated before the search equality function is applied. In others, such as JavaScript, an .indexOf search will not match a string like "epsilon" with a potentially equivalent function call like (function () {return "epsilon"})() It might be helpful to specify requirements which drew these things out and made them visible. Hout (talk) 13:13, 10 October 2015 (UTC)

A couple of concrete examples – I think the requirements should at least prompt, in the case of JavaScript ES6, the supply of a specific equality function by the coder, returning a value in the case of Array.prototype.find(callback), and an index in the case of Array.prototype.findIndex(callback).

For JS ES5, it should prompt the use/demonstration of coder-specified equality functions through equivalent 'polyfills'

In the case of Haskell, the requirements should be such as to require the supply of a predicate function to find :: (a -> Bool) -> [a] -> Maybe a

and also to findIndex :: (a -> Bool) -> [a] -> Maybe Int

Hout (talk) 13:33, 10 October 2015 (UTC)

That sounds plausible. Please feel free to update the task description - it's not like you have to worry about messing that up... --Rdm (talk) 13:40, 10 October 2015 (UTC)

But on second thoughts, do we need this task anyway, given the existing http://rosettacode.org/wiki/Search_a_list ? Hout (talk) 14:43, 10 October 2015 (UTC)

That's good question. One relevant issue might be that Search a list specifies an index result, while this task allows non-index results. --Rdm (talk) 05:27, 11 October 2015 (UTC)

A name change here, perhaps ? -> sth like 'Find the first match in a vector' ? Hout (talk) 06:44, 11 October 2015 (UTC)

Population exceeding or population less than ? this is confusing --G.Brougnard (talk) 10:10, 11 October 2015 (UTC)

First item with population below a certain limit. (less than X) (assumes that this list is sorted (as shown) in descending size of population, and that we are searching from the largest city at the beginning towards the smallest city at the end). Is there an alternative wording that you might prefer ? Hout (talk) 20:21, 11 October 2015 (UTC)

Maybe add find all city's with pop < 5 million or some such to get multiple results --CraigD (talk) 19:58, 11 October 2015 (UTC)

I wonder if that would take us from the simplest 'first match' of functions like JS's .indexOf(), .find(), and .findIndex(), into a broader database query ? Perhaps there's an argument for restricting the scope a little ? Not sure … Hout (talk) 20:21, 11 October 2015 (UTC)