www.digitalmars.com         C & C++   DMDScript  

digitalmars.D - Reactor, fiber scheduler and async in Ruby 3.0

reply Jacob Carlborg <doob me.com> writes:
In Ruby 3.0, support for reactor, fiber scheduler and async has been added.

async seems to be implemented in library code. The existing IO related 
methods in the core/standard library has been updated to be fiber 
scheduler aware. That basically means that if there is no fiber 
scheduler set, the IO operations will be blocking. If a fiber scheduler 
has been set, they will instead be non-blocking. A lot of existing code 
and gems will just work without having to be adopted for async.

Here's an example:

require 'async'
require 'net/http'
require 'uri'

Async do
   ["ruby", "rails", "async"].each do |topic|
     Async do
       Net::HTTP.get(URI "https://www.google.com/search?q=#{topic}")
     end
   end
end

IO#read and IO#write are two of the methods that have been made fiber 
scheduler aware. Net::HTTP#get uses these methods under the hood and 
therefore will be non-blocking inside the Async block. See [1] for more 
details.

BTW, Zig uses the same idea with the underlying IO functions that can 
adopt and become non-blocking.

https://www.ruby-lang.org/en/news/2020/12/25/ruby-3-0-0-released/

-- 
/Jacob Carlborg
Dec 31 2020
next sibling parent reply Adam D. Ruppe <destructionator gmail.com> writes:
On Thursday, 31 December 2020 at 14:53:52 UTC, Jacob Carlborg 
wrote:

 The existing IO related methods in the core/standard library 
 has been updated to be fiber scheduler aware. That basically 
 means that if there is no fiber scheduler set, the IO 
 operations will be blocking. If a fiber scheduler has been set, 
 they will instead be non-blocking.


omg that's a really good idea.

Over the weekend I wrote a Phobos subclass 
<http://dpldocs.info/experimental-docs/arsd.fibersocket.html> 
that does fiber sockets. Pretty simple code really.

But the connect* family of functions just assert they are inside 
a fiber right now.... but since they implement the same Socket 
interface, I could just do `if(fiber.getThis is null) return new 
Socket; else return new FiberSocket` and do the same behavior 
ruby is showing.

Very convenient...

(the difference between this and ruby though is i still have a 
FiberManager class you use as the socket factory, whereas I 
suppose they just made the existing methods return appropriate 
classes. Or heck I could even just make the overridden methods be 
like `if(thisFiber is null) super.receive(); else { ... }`. 
That'd be reasonably acceptable too. Heck even Phobos itself 
could potentially do that.)
Dec 31 2020
parent ikod <igor.khasilev gmail.com> writes:
On Thursday, 31 December 2020 at 16:32:17 UTC, Adam D. Ruppe 
wrote:

 On Thursday, 31 December 2020 at 14:53:52 UTC, Jacob Carlborg 
 wrote:

 The existing IO related methods in the core/standard library 
 has been updated to be fiber scheduler aware. That basically 
 means that if there is no fiber scheduler set, the IO 
 operations will be blocking. If a fiber scheduler has been 
 set, they will instead be non-blocking.


 omg that's a really good idea.

 Over the weekend I wrote a Phobos subclass 
 <http://dpldocs.info/experimental-docs/arsd.fibersocket.html> 
 that does fiber sockets. Pretty simple code really.


Yes, this is how I implemented this in my event loop framework:
https://github.com/ikod/hio/blob/master/source/hio/socket/package.d#L1275

Then your application can just ignore details of underlying 
socket implementation (it can be std.socket or unblocking sockets 
with event loop)
Dec 31 2020
prev sibling parent reply James Blachly <james.blachly gmail.com> writes:
On 12/31/20 9:53 AM, Jacob Carlborg wrote:

 In Ruby 3.0, support for reactor, fiber scheduler and async has been added.
 
 async seems to be implemented in library code. The existing IO related 


...

Marginally on-topic:

Is Photon dead?

I understand Weka are no longer updating Mecca, yes?
Jan 02 2021
next sibling parent Jacob Carlborg <doob me.com> writes:
On 2021-01-03 03:50, James Blachly wrote:


 I understand Weka are no longer updating Mecca, yes?


Yes, it has not been updated in 1.5 years. I asked in Slack about the 
future plans, I got this answer: "There's no formal answer to that, yet. 
I'll try and find out".

-- 
/Jacob Carlborg
Jan 06 2021
prev sibling next sibling parent reply Dmitry Olshansky <dmitry.olsh gmail.com> writes:
On Sunday, 3 January 2021 at 02:50:27 UTC, James Blachly wrote:

 On 12/31/20 9:53 AM, Jacob Carlborg wrote:

 In Ruby 3.0, support for reactor, fiber scheduler and async 
 has been added.
 
 async seems to be implemented in library code. The existing IO 
 related




[snip]


 Marginally on-topic:

 Is Photon dead?


Mostly dead, yes, my spare time is very limited these days, so 
not doing much of anything in the D world.

The biggest issue in reviving say Photon is the fact that I do 
not have a good project in mind to benefit from the 
infrastructure code of Photon.


—
Dmitry Olshansky
Jan 18 2021
parent James Blachly <james.blachly gmail.com> writes:
On 1/18/21 2:21 PM, Dmitry Olshansky wrote:

 On Sunday, 3 January 2021 at 02:50:27 UTC, James Blachly wrote:

 Marginally on-topic:

 Is Photon dead?


 
 Mostly dead, yes, my spare time is very limited these days, so not doing 
 much of anything in the D world.
 
 The biggest issue in reviving say Photon is the fact that I do not have 
 a good project in mind to benefit from the infrastructure code of Photon.
 
 
 —
 Dmitry Olshansky


Understandable, and thank you for what you have done; in particular, 
showing what is possible.
Jan 19 2021
prev sibling parent reply Jacob Carlborg <doob me.com> writes:
On Sunday, 3 January 2021 at 02:50:27 UTC, James Blachly wrote:


 I understand Weka are no longer updating Mecca, yes?


Mecca has now been move to dlang-community group [1]. I intend to 
at least make sure it works with the latest compilers.

[1] https://github.com/dlang-community/mecca

--
/Jacob Carlborg
Jan 19 2021
next sibling parent Imperatorn <johan_forsberg_86 hotmail.com> writes:
On Tuesday, 19 January 2021 at 15:26:34 UTC, Jacob Carlborg wrote:

 On Sunday, 3 January 2021 at 02:50:27 UTC, James Blachly wrote:


 I understand Weka are no longer updating Mecca, yes?


 Mecca has now been move to dlang-community group [1]. I intend 
 to at least make sure it works with the latest compilers.

 [1] https://github.com/dlang-community/mecca

 --
 /Jacob Carlborg


🍀
Jan 19 2021
prev sibling parent James Blachly <james.blachly gmail.com> writes:
On 1/19/21 10:26 AM, Jacob Carlborg wrote:

 On Sunday, 3 January 2021 at 02:50:27 UTC, James Blachly wrote:
 

 I understand Weka are no longer updating Mecca, yes?


 
 Mecca has now been move to dlang-community group [1]. I intend to at 
 least make sure it works with the latest compilers.
 
 [1] https://github.com/dlang-community/mecca
 
 -- 
 /Jacob Carlborg


unicode-beer-glyph =)
Jan 19 2021