I would like to read from a pipe in which data may arrive slowly. From
experimenting, it looks like os.read() will block until it returns the
maximum amount of data you asked for, in the second parameter to read(),
or until it hits EOF. I cannot find a way to return only the data that
the file object has immediately available.

If no data is available, blocking is OK.

The only workaround I can think of is to call select() in a loop,
reading and storing one byte each time, and then returning them when
select() indicates that the pipe is not ready for reading.

Are there better approaches? Thanks,

-Peter

In article <cnbgd0$ls4$1@news.apple.com>,
Peter Ammon <peter_ammon@rocketmail.com> wrote:
[color=blue]
> I would like to read from a pipe in which data may arrive slowly. From
> experimenting, it looks like os.read() will block until it returns the
> maximum amount of data you asked for, in the second parameter to read(),
> or until it hits EOF. I cannot find a way to return only the data that
> the file object has immediately available.
>
> If no data is available, blocking is OK.
>
> The only workaround I can think of is to call select() in a loop,
> reading and storing one byte each time, and then returning them when
> select() indicates that the pipe is not ready for reading.[/color]

I don't think that will work either!

But os.read (a.k.a. posix.read) does just what you want.
Use it with a file descriptor -

fp = os.popen(cmd, 'r')
fd = fp.fileno()
while 1:
data = os.read(fd, 4096)
if not data:
break
dispose_of(data)

Don't use the file object read or readline methods at all,
because they could leave data in their own input buffer,
where posix.read can't see it (nor select, which is why
I doubt your work-around would work.)

Donn Cave, donn@u.washington.edu