[Caml-list] [ANN] Hmap 0.8.0

[Caml-list] [ANN] Hmap 0.8.0

[Daniel Bünzli]

Hello, 

I'd like to announce the first release of Hmap: 

```
Hmap provides heterogeneous value maps for OCaml. These maps bind keys
to values with arbitrary types. Keys witness the type of the value
they are bound to which allows to add and lookup bindings in a type
safe manner.
```

Hmap has no dependency and is distributed under the ISC license.

Home page: http://erratique.ch/software/hmap/
API docs: http://erratique.ch/software/hmap/doc/Hmap

Best, 

Daniel

Re: [Caml-list] [ANN] Hmap 0.8.0

[Yaron Minsky]

[-- Attachment #1: Type: text/plain, Size: 1349 bytes --]

Neat!

I'm curious if you've looked at Core_kernel's Univ_map?  I believe it's
very similar in spirit, and I'm curious if there are any interesting
differences there:

https://github.com/janestreet/core_kernel/blob/master/src/univ_map.mli

Our type witnesses are a little interesting in that you can derive new
witnesses from old ones (e.g., if a equals a', and b equals b', then a * b
equals a' * b').  That said, I'm not sure that this is relevant for the
heterogenous map use case.

y

On Tue, Mar 8, 2016 at 4:42 AM, Daniel Bünzli <daniel.buenzli@erratique.ch>
wrote:

> Hello,
>
> I'd like to announce the first release of Hmap:
>
> ```
> Hmap provides heterogeneous value maps for OCaml. These maps bind keys
> to values with arbitrary types. Keys witness the type of the value
> they are bound to which allows to add and lookup bindings in a type
> safe manner.
> ```
>
> Hmap has no dependency and is distributed under the ISC license.
>
> Home page: http://erratique.ch/software/hmap/
> API docs: http://erratique.ch/software/hmap/doc/Hmap
>
> Best,
>
> Daniel
>
> --
> Caml-list mailing list.  Subscription management and archives:
> https://sympa.inria.fr/sympa/arc/caml-list
> Beginner's list: http://groups.yahoo.com/group/ocaml_beginners
> Bug reports: http://caml.inria.fr/bin/caml-bugs
>

[-- Attachment #2: Type: text/html, Size: 2417 bytes --]

Re: [Caml-list] [ANN] Hmap 0.8.0

[Daniel Bünzli]

Le mardi, 8 mars 2016 à 13:44, Yaron Minsky a écrit :
> I'm curious if you've looked at Core_kernel's Univ_map?
No, I just abstracted code I've been c&p around in a lot of projects.

> Our type witnesses are a little interesting in that you can derive new witnesses from old ones (e.g., if a equals a', and b equals b', then a * b equals a' * b'). That said, I'm not sure that this is relevant for the heterogenous map use case.
The type identifiers are an implementation detail. I used to implement this using a universal type, see e.g. [1], but the type identifier implementation allocates less and should be more efficient (but the latter I didn't measure).  

Best,  

Daniel

[1] http://alan.petitepomme.net/cwn/2010.02.09.html#1