On Nov 2, 2013, at 9:27 AM, Mark Engelberg <[email protected]> wrote:
> I seem to be a relatively lone voice thinking it makes sense to preserve the
> type of the map, and honestly, I'm not going to lose any sleep if this patch
> is never actually implemented, but here's my two cents on the "optimal"
> design for records.
>
> Earlier in the thread, I mentioned two mental models about how select-keys
> works. Even though the actual implementation involves starting from an empty
> map and building back up, I think the behavior of select-keys should mimic
> the mental model of discarding irrelevant keys.
>
> The two mental models predict somewhat differently what would happen when
> working with records. Records can hold additional keys. So imagine someone
> dumps a whole lot of additional keys, and then wants to "recover" the
> original, true record, by calling select-keys with the original keys in the
> record. I think this is a use case worth supporting. This corresponds to
> the mental model of discarding unwanted keys.
>
> To implement this for records:
> If any of the original record keys do not appear in the selected keys, then,
> as with dissoc, we must revert back to a map as there is no valid way to
> represent this as a record.
> Otherwise, maintain the record type -- the core record contents are
> preserved, and essentially you only need to call select-keys only on the
> extended map of extra keys stored in the record.
I agree 100% with Mark’s comments above.
This is exactly the model I would expect select-keys to have.
Conceptually, this is how I imagine select-keys should work:
(defn select-keys++
"Returns a new map of the same (hashed/sorted) type,
that contains only those entries whose key is in keyseq"
[map keyseq]
(apply dissoc
map
(clojure.set/difference (set (keys map))
(set keyseq))))
I am not saying that this would be the best , or even a good, implementation.
Like Mark, I am not going to lose any sleep about this either, but I was using
Mark’s data.priority-maps the other day,
did a select-keys to winnow down my priority-map, and was surprised by the
result.
Perhaps I should not have been surprised….
In preparing this email, I carefully re-read the doc strings for both dissoc
and select keys:
(defn dissoc
"dissoc[iate]. Returns a new map of the same (hashed/sorted) type,
that does not contain a mapping for key(s).”
(defn select-keys
"Returns a map containing only those entries in map whose key is in keys"
Clearly dissoc makes the promise to return a map of the same type that
select-keys doesn’t.
A potential non-code change to select-keys might be to emphasize that the
return map will be
a hash-map, regardless of the kind/type of input map.
Don
>
>
> On Sat, Nov 2, 2013 at 8:49 AM, Andy Fingerhut <[email protected]>
> wrote:
> I attached another patch to the ticket. It builds up the answer from the
> empty map {} if the argument is a record (as the current select-keys does),
> but from (empty map) if it is not a record, so it will preserve sortedness of
> the argument. Not sure if there are any other cases that are a problem, or
> if it does what everyone would expect, but it should be closer.
>
> http://dev.clojure.org/jira/browse/CLJ-1287
>
> Andy
>
>
> On Sat, Nov 2, 2013 at 6:07 AM, Alex Miller <[email protected]> wrote:
> > One other point:
> > Sometimes people use sorted maps and array maps specifically for scenarios
> > in which the keys are not hashable and therefore hash maps would not apply.
> > Dumping the contents into a regular map in such cases doesn't make much
> > sense.
>
> Everything is hashable, not sure what a non-hashable key means. Array maps
> use the hash of the key to determine the array bucket. If you get the hash
> code of a sorted map, it will get the hash of all keys and values.
>
> --
> --
> You received this message because you are subscribed to the Google
> Groups "Clojure" group.
> To post to this group, send email to [email protected]
> Note that posts from new members are moderated - please be patient with your
> first post.
> To unsubscribe from this group, send email to
> [email protected]
> For more options, visit this group at
> http://groups.google.com/group/clojure?hl=en
> ---
> You received this message because you are subscribed to the Google Groups
> "Clojure" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to [email protected].
> For more options, visit https://groups.google.com/groups/opt_out.
>
>
> --
> --
> You received this message because you are subscribed to the Google
> Groups "Clojure" group.
> To post to this group, send email to [email protected]
> Note that posts from new members are moderated - please be patient with your
> first post.
> To unsubscribe from this group, send email to
> [email protected]
> For more options, visit this group at
> http://groups.google.com/group/clojure?hl=en
> ---
> You received this message because you are subscribed to the Google Groups
> "Clojure" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to [email protected].
> For more options, visit https://groups.google.com/groups/opt_out.
>
>
> --
> --
> You received this message because you are subscribed to the Google
> Groups "Clojure" group.
> To post to this group, send email to [email protected]
> Note that posts from new members are moderated - please be patient with your
> first post.
> To unsubscribe from this group, send email to
> [email protected]
> For more options, visit this group at
> http://groups.google.com/group/clojure?hl=en
> ---
> You received this message because you are subscribed to the Google Groups
> "Clojure" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to [email protected].
> For more options, visit https://groups.google.com/groups/opt_out.
--
--
You received this message because you are subscribed to the Google
Groups "Clojure" group.
To post to this group, send email to [email protected]
Note that posts from new members are moderated - please be patient with your
first post.
To unsubscribe from this group, send email to
[email protected]
For more options, visit this group at
http://groups.google.com/group/clojure?hl=en
---
You received this message because you are subscribed to the Google Groups
"Clojure" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
For more options, visit https://groups.google.com/groups/opt_out.
