[ann] fibers 1.0.0 released

[ann] fibers 1.0.0 released

[Andy Wingo]

Hello!

I'm delighted to announce the release of Fibers 1.0.0.

Fibers is a lightweight concurrency facility for Guile that supports
non-blocking input and output, millions of concurrent threads, and
Concurrent ML-inspired communication primitives.  For more information,
see the web version of the manual at:

  https://github.com/wingo/fibers/wiki/Manual

The manual is also available in PDF and Info formats in the tarball.

The 1.0.0 tarball can be found here:

  https://wingolog.org/pub/fibers/fibers-1.0.0.tar.gz

Its SHA256 sum is:

  a051913e252f872c969d71939310cee10844157a317db93ebcaf3df8c479536e  fibers-1.0.0.tar.gz

Fibers 1.0.0 requires the newly released Guile 2.1.7.

This release finally improves the exception handling situation, so that
that fibers that have uncaught exceptions print nice errors instead of
killing the system.  It also fixes a bug in channels, adds a condition
variable implementation, adds a standalone web server implementation
that scales with the number of cores available to it, implements many
performance improvements, lowers the overhead of preemption, and
improves documentation.

Happy hacking,

Andy

Re: [ann] fibers 1.0.0 released

[Matt Wette]

> On Feb 20, 2017, at 12:19 PM, Andy Wingo <wingo@pobox.com> wrote:
> 
> Hello!
> 
> I'm delighted to announce the release of Fibers 1.0.0.

Sweet!  I have read through Section 1 and I now have another tool in the box. Nice effort.  — Matt

Re: [ann] fibers 1.0.0 released

[Amirouche]

Le 20/02/2017 à 22:06, Matt Wette a écrit :
>> On Feb 20, 2017, at 12:19 PM, Andy Wingo <wingo@pobox.com> wrote:
>>
>> Hello!
>>
>> I'm delighted to announce the release of Fibers 1.0.0.
> Sweet!  I have read through Section 1 and I now have another tool in the box. Nice effort.  — Matt

+1

Re: [ann] fibers 1.0.0 released

[Matt Wette]

> On Feb 20, 2017, at 12:19 PM, Andy Wingo <wingo@pobox.com> wrote:
> 
> Hello!
> 
> I'm delighted to announce the release of Fibers 1.0.0.
> 
> Fibers is a lightweight concurrency facility for Guile that supports
> non-blocking input and output, millions of concurrent threads, and
> Concurrent ML-inspired communication primitives.  For more information,
> see the web version of the manual at:
> 
>  https://github.com/wingo/fibers/wiki/Manual
> 
> The manual is also available in PDF and Info formats in the tarball.
> 
> The 1.0.0 tarball can be found here:
> 
>  https://wingolog.org/pub/fibers/fibers-1.0.0.tar.gz
> 
> Its SHA256 sum is:
> 
>  a051913e252f872c969d71939310cee10844157a317db93ebcaf3df8c479536e  fibers-1.0.0.tar.gz
> 
> Fibers 1.0.0 requires the newly released Guile 2.1.7.
> 
> This release finally improves the exception handling situation, so that
> that fibers that have uncaught exceptions print nice errors instead of
> killing the system.  It also fixes a bug in channels, adds a condition
> variable implementation, adds a standalone web server implementation
> that scales with the number of cores available to it, implements many
> performance improvements, lowers the overhead of preemption, and
> improves documentation.
> 
> Happy hacking,
> 
> Andy
> 
> 

This is not going to compile on macOS or FreeBSD, but these OSes do support kqueue.  
How to proceed?  I think making equeue look like epoll may be a bit kludgy.

And the cpuclockid calls need to be dealt with:
* clock_getcpuclockid
* pthread_getcpuclockid
* clock_nanosleep

Matt

Re: [ann] fibers 1.0.0 released

[Matt Wette]

> On Oct 31, 2017, at 7:31 PM, Matt Wette <matt.wette@gmail.com> wrote:
>> On Feb 20, 2017, at 12:19 PM, Andy Wingo <wingo@pobox.com> wrote:
>> 
>> Hello!
>> 
>> I'm delighted to announce the release of Fibers 1.0.0.
> 
> This is not going to compile on macOS or FreeBSD, but these OSes do support kqueue.  
> How to proceed?  I think making equeue look like epoll may be a bit kludgy.
> 
> And the cpuclockid calls need to be dealt with:
> * clock_getcpuclockid
> * pthread_getcpuclockid
> * clock_nanosleep

I wil take a shot at initial port to kqueue.  If you have no interest let me know.
I am starting by working internals.scm to see if I can isolate epoll dependencies.

Matt