Re^4: Stumped with select->can_read (buffered)

[w1r3d]

in reply to Re^3: Stumped with select->can_read (buffered)
in thread Stumped with select->can_read

I tried doing that, but it didn't seem to work. My guess is that because one particular $fh is very busy, the while loop never ends?

This *seems* to work at the moment:

while(@ready = $sel->can_read)
{
  foreach my $fh(@ready)
  {
    my $line = "";
    sysread( $fh, $line, 1024 );
  }
}

My concern now is whether 1024 is too small of a buffer that I could miss something (if that's possible), or missing a newline character because of the buffer size. Would it be horrible (performance wise) to read one byte at a time until I reach a newline character, thus forming a real $line?

Thanks again!

Pedro

[tye]

My concern now is whether 1024 is too small of a buffer that I could miss something (if that's possible)

If there are more than 1024 bytes to be read, then the next call to can_read() will tell you about them.

[zentara]

Would it be horrible (performance wise) to read one byte at a time until I reach a newline character, thus forming a real $line?

It's been done before. :-) It certainly is inefficient. What I would do, is sysread as big of chunks as you can, then concat everything into a temp buffer, then split the buffer on newlines into an @array. If the scalar @array is greater than 1, then you have a line. Return the last @array element to your $temp_buffer, and process the rest of the @array elements as lines. It's called a sliding buffer, and you can google for "perl sliding buffer" for examples.

[tye]

sysread as big of chunks as you can, then concat everything into a temp buffer

sysread can read onto the end of the buffer, avoiding the need to copy the data again:

    sysread( $fh, $buf, 16*1024, length($buf) );