RFR(s): 8244733: Add ResourceHashtable::xxx_if_absent (was: ResourceHashtable::compute_if_absent)

[coleen.phillimore at oracle.com]

I just spent time looking for a newIfAbsent in java.util.Hash, but there 
isn't one.  These three functions are eye-strainingly similar.

Can you remove new_if_absent and have another compute_if_absent(key, 
created) overload that calls

+ V* compute_if_absent(K const& key, mapping_fun_t mapping_function, 
bool* p_created = NULL) {


with a trivial mapping function, like your test.  And remove the default 
parameters since I think none of the callers should have a default 
parameter.

thanks,
Coleen

On 5/14/20 4:08 AM, Thomas Stüfe wrote:
> Hi,
>
> This is a continuation of the review thread:
> https://mail.openjdk.java.net/pipermail/hotspot-runtime-dev/2020-May/039444.html
>
> I changed the title of the bug somewhat hence the new thread.
>
> JBS: https://bugs.openjdk.java.net/browse/JDK-8244733
> Webrev:
> http://cr.openjdk.java.net/~stuefe/webrevs/8244733--add-resourcehashtable--compute_if_absent/webrev.01/webrev/
>
> This now adds three new functions to ResourceHashTable:
>
> new_if_absent(), compute_if_absent(), put_if_absent()
>
> which work very similar to their equivalents in j.u.Map.
>
> The advantage they offer is that for a typical "if map does not contain X,
> add X+value" pattern we only do the lookup once.
>
> --
>
> So I extended the scope of the item. After playing around with various call
> sites I realized that David's proposal of put_if_absent() would be useful,
> as well as a variant which just adds a default constructed value. All three
> are tested in gtest and used in real call sites.
>
> Ran tests tonight in our CI, all went through.
>
> Thanks, Thomas