[ast-developers] SIGPOLL support?

[Lionel Cons]

Is anyone here interested in SIGPOLL support in ksh93 (after the
.sh.sig support has been fixed), i.e. provide a way to open a fd with
O_ASYNC and listen to the SIGIO or SIGPOLL signals.

Lionel

[Cedric Blancher]

Post by Lionel Cons
Is anyone here interested in SIGPOLL support in ksh93 (after the
.sh.sig support has been fixed), i.e. provide a way to open a fd with
O_ASYNC and listen to the SIGIO or SIGPOLL signals.

I assume you know about the poll(1) builtin ksh93 now supports.

What's the purpose of SIGPOLL support? Trying to implement push
notification for socket I/O instead of pull via poll(1)?

Ced

[Lionel Cons]

Post by Cedric Blancher

Post by Lionel Cons
Is anyone here interested in SIGPOLL support in ksh93 (after the
.sh.sig support has been fixed), i.e. provide a way to open a fd with
O_ASYNC and listen to the SIGIO or SIGPOLL signals.

I assume you know about the poll(1) builtin ksh93 now supports.

Yes, sure. We use poll almost everywhere now

Post by Cedric Blancher
What's the purpose of SIGPOLL support? Trying to implement push
notification for socket I/O instead of pull via poll(1)?

Yes, albeit its useless to discuss or even try to implement until
signals work reliably and the traps are called in the order the
signals arrived.

Lionel

[Cyrille Lefevre]

Post by Cedric Blancher

Post by Lionel Cons
Is anyone here interested in SIGPOLL support in ksh93 (after the
.sh.sig support has been fixed), i.e. provide a way to open a fd with
O_ASYNC and listen to the SIGIO or SIGPOLL signals.

I assume you know about the poll(1) builtin ksh93 now supports.
What's the purpose of SIGPOLL support? Trying to implement push
notification for socket I/O instead of pull via poll(1)?

instead of make an active loop to see if there is something to read on
some fd, you may do anything you want until a SIGIO trap raise some
handler which know that there is something to read.

c'est plus clair maintenant ? :-)

Regards,

Cyrille Lefevre