Greetings, esteemed monks!

So I have been tilting at the windmill of handling STDOUT and STDERR of a child
synchronously, yet separately. I work in a Linux environment (RedHat 3 and 4 and perl 5.8.5).
I've resigned myself to getting halfway results with
sporadic attempts at refining my technique. I won't repeat myself here, but this node:

Re: Synchronizing STDERR and STDOUT

explains the current status of my efforts.

Here's an example of some output from the example child program:

<snip>
<span id='stderr'>1</span>
2
<span id='stderr'>3</span>
4
<span id='stderr'>5</span>
6
8
10
12
14
16
18
<span id='stderr'>7
9
11
13
15
17
19
21
23
25
27
29
31
33
35</span>
20
<snip>
58
<span id='stderr'>37
39
<snip>
73</span>
60
62
<you get the picture>
[download]

As you can see in the code, autoflushing is turned on (in both child and parent) and the child should think it's connected to a tty.
However, the output is still buffered in chunks larger than one line.

Upon reading the whole thread that the above node is in, I found out about PerlIO and enabled logging for a run of the program.

I glanced through perlio.log and I found the following line:

/opt/perl/lib/site_perl/5.8.5/IPC/Run.pm:2613 Cannot flush f=0x805defc

here is line 2613 in my version of Run.pm:

open $s1, ">&=$self->{SYNC_WRITER_FD}"
        or croak "$! setting filehandle to fd SYNC_WRITER_FD" ;
[download]

I am curious about where the "Cannot flush" message originated and why. Any ideas?

Terrence

_________________________________________________________________________________

I like computer programming because it's like Legos for the mind.