[Caml-list] One more question about Gc.finalise

[Caml-list] One more question about Gc.finalise

[Evgeny Roubinchtein]

[-- Attachment #1: Type: text/plain, Size: 538 bytes --]

Dear OCaml users and developers,

I have one more question about Gc.finalise to which I couldn't find an
answer in the documentation.
Suppose I do, "Gc.finalise f v", and the function f makes the value v
reachable again.  If I want f to be called again should v become
unreachable again, do I need to again call "Gc.finalise f v", after f has
"revived" v?

My experimentation _seems_ to suggest that I do, but there is always the
danger that I am misinterpreting the results of my experiment. :-)

Thank you in advance!

-- 
Best,
Zhenya

[-- Attachment #2: Type: text/html, Size: 701 bytes --]

Re: [Caml-list] One more question about Gc.finalise

[François Bobot]

Hi,

Le 19/06/2017 à 01:45, Evgeny Roubinchtein a écrit :
> I have one more question about Gc.finalise to which I couldn't find an answer in the documentation.
> Suppose I do, "Gc.finalise f v", and the function f makes the value v reachable again.  If I want f
> to be called again should v become unreachable again, do I need to again call "Gc.finalise f v",
> after f has "revived" v?


I think so, but I didn't double checked with the code.

>
> My experimentation _seems_ to suggest that I do, but there is always the danger that I am
> misinterpreting the results of my experiment. :-)


With newer ocaml version (>= 4.04), you can use `GC.finalise_last` which can't use the value 
finalised (unit instead of 'a is given to the function `f`) but the function `f` will really be 
called after `v` is unreachable for the last time, just before being reclaimed. Perhaps it could fit 
better what you have in mind.

Best,

-- 
François

Re: [Caml-list] One more question about Gc.finalise

[Jacques-Henri Jourdan]

[-- Attachment #1: Type: text/plain, Size: 655 bytes --]



On 06/19/2017 08:02 AM, François Bobot wrote:
> Hi,
>
> Le 19/06/2017 à 01:45, Evgeny Roubinchtein a écrit :
>> I have one more question about Gc.finalise to which I couldn't find 
>> an answer in the documentation.
>> Suppose I do, "Gc.finalise f v", and the function f makes the value v 
>> reachable again.  If I want f
>> to be called again should v become unreachable again, do I need to 
>> again call "Gc.finalise f v",
>> after f has "revived" v?
>
>
> I think so, but I didn't double checked with the code. 

Indeed it is. This is even used to implement alarm in stdlib :

https://github.com/ocaml/ocaml/blob/trunk/stdlib/gc.ml#L103

-- 
JH

[-- Attachment #2: Type: text/html, Size: 1363 bytes --]