apr_pollset_wakeup() on strange(ish) platforms

[Branko Čibej <br...@apache.org>]

Hi,

I'm looking at the implementation of apr_pollset_wakeup(), and if I'm 
reading this correctly, if APR_FILES_AS_SOCKETS is 0, we only implement 
funciton for Windows and OS/2. All other platforms just return APR_ENOTIMPL.

But. Is there any reason why there couldn't be a generic implementation 
for these other platforms, using a socket chasing its own tail on the 
loopback interface, or some platform-specific local domain socket (the 
OS/2 implementation does this)?

I realise the platforms that are currently not covered are not exactly 
major, but it I also can't think of a reason for this incompleteness.

-- Brane

Re: apr_pollset_wakeup() on strange(ish) platforms

[Eric Covener <co...@gmail.com>]

On Fri, Jul 18, 2025 at 11:06 AM Branko Čibej <br...@apache.org> wrote:
>
> Hi,
>
> I'm looking at the implementation of apr_pollset_wakeup(), and if I'm reading this correctly, if APR_FILES_AS_SOCKETS is 0, we only implement funciton for Windows and OS/2. All other platforms just return APR_ENOTIMPL.
>
> But. Is there any reason why there couldn't be a generic implementation for these other platforms, using a socket chasing its own tail on the loopback interface, or some platform-specific local domain socket (the OS/2 implementation does this)?

> I realise the platforms that are currently not covered are not exactly major, but it I also can't think of a reason for this incompleteness.

I was looking at this a year ago or so. It does seems like replacing
the Windows-isms with the basic socket APIs would be pretty
safe/portable.