Re: [Chicken-users] replace signal with sigaction

[Jörg F . Wittenberger]

On Sep 30 2011, Jörg F. Wittenberger wrote:

> Alan,
>
> I figured that you did almost have you for the process-io-ports code
> I cited the other day.

Alan, another warning:  I just fixed a case of EINTR, where the code
shows strong heritage from the tcp unit.  (I wrote some parallel code
base on it, because the API did not fit my needs.  At the low level the
code is written, establishing a signal handler around each call just to
catch the highly probable -1 exception would have made a mess out of it.
Therefore I'm calling the accept(2) etc. via Chicken's FFI and handle all
the results in one place.)

Nevertheless, the basic structure of the code is the same.  So I'm afraid
that there are more occurrences of unhandled EINTR in the tcp unit.
So if your code depends on it, be prepared.

/Jörg