Hi, Thank you George for your answer.
On Wednesday, October 13, 2021 at 5:37:39 PM UTC+2 gneuner2 wrote: > > On 10/12/2021 7:01 PM, unlimitedscolobb wrote: > > I wrote myself this little function: > > > > (define (hash->ordered-list h) > > (hash-map h cons #t)) > > > > which uses the try-order? argument of hash-map. > > > > Is there a reason for hash->list not have an optional argument > > try-order? Or perhaps having such a standalone function would be > better? > > > > I was planning to submit a patch, but then I thought I may be missing > > something. > > > > - > > Sergiu > > > > I can't speak for the Racket team, but ... > > Hash tables entries inherently are unordered, so there really is no > reason to expect an ordered list from reading out the data. Also, the > docs indicate that 'try-order?' doesn't work for all data types - so it > may produce unexpected results. I have two main use cases for producing an ordered list from a hash table: 1. A canonical way to pretty print a hash table: In my projects, I carry around and print out hash tables a lot, so I like the elements to appear in the same order all the time, whatever that order may be. Luckily, `hash-map` orders symbols alphabetically and numbers according to the natural order, so it's perfect for my use. 2. Testing the contents of a mutable hash table: The only way I found to that is to convert the hash table to an ordered list and compared it to the test list. This is clearly not the most efficient use of a hash table, but I can totally go with that, since it's about testing and not the actual performance. Of course, I am totally open to learning better ways of doing these things! > Further, sorting the output potentially > can take a lot of extra time ... having never tried to get ordered > output from hash-map, I can only hypothesize that (at least in some > cases) it may be faster to create the unordered list and then sort it > than to try to produce an ordered list with 'try-order?'. > I scrounged around Racket sources a bit, and even though I am not sure I looked in the right place, I got the impression that that is what hash-map with try-order? set to #t does already: creating an unordered list out of the hash table and then `sort`ing it. Certainly, you are welcome to submit a change, but I think it would be > best to leave the existing behavior and make ordered output an addition. > > FYI, here's the PR: https://github.com/racket/racket/pull/4025 I pored a lot over this, and the reasoning which won at the end was that `hash->list` is basically a call to `hash-map` with `cons` as the function to apply. Thus, adding try-order? to hash->list is a very easy change, so why not have it. Just to be clear, my PR does not alter the default behavior of `hash->list`, and only adds an optional argument. - Sergiu -- You received this message because you are subscribed to the Google Groups "Racket Users" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To view this discussion on the web visit https://groups.google.com/d/msgid/racket-users/a1e0bbd2-69f8-439e-b362-306c897926d3n%40googlegroups.com.
