From: Zach Dennis <zdennis@...>
Date: 2005-02-08T00:52:58+09:00
Subject: Re:

Caleb,

This message showed up under another, thread perhaps you replied to a 
message and just changed the subject heading. This makes it difficult 
for certain mail clients and newsreaders to correctly organize the mail 
in threads, next time could you compose a new email messsage to 
"ruby-talk@ruby-lang.org", that way more people will see your email, 
otherwise people may ignore the thread that this messages shows up under.

thx,

Zach



Caleb Tennis wrote:
> I've got a simple class that writes to a device on a serial port and then
> reads back a response:
> 
> class SerialPort
>   def write(port, arry)
>     file = File.new(port, File::RDWR | File::NOCTTY )
>     file.write(arry.pack("C*"))
>     file.flush
>     str = file.read(50).to_s.unpack("C*")
>     file.close
>     str
>   end
> end
> 
> (Usage:  response = SerialPort.new.write("/dev/tts/5",
> [some_array_of_bytes]) )
> 
> The class works fine for my program, until I go multi-threaded, which is a
> requirement.
> 
> As soon as I do something this:
> 
> Thread.new do
>   response = SerialPort.new.write("/dev/tts/5, [some_array_of_bytes])
> end
> 
> The thread hangs at the read() function.  If I use the same code in the
> main thread it works fine.  I thought at first I had resources that were
> conflicting, but I've written a test case which does nothing other than
> one call like seen above, and the only difference in working/not-working
> is whether the code is invoked in a new thread or not.
> 
> Any thoughts on what I might be missing here?  The serial port is set to
> raw mode, min 0 time 1, which means that it won't block on reads even if
> no data is present.
> 
> 
> 
>