On 25/11/05, Bruce McKenzie <[EMAIL PROTECTED]> wrote: > Hi. > > I've written an agent (using Mech) to search for audiobooks and eAudio > books in several libraries. As this takes quite a while, I'd like to be > able to make the browser display progress results. > > I can launch the agent and then launch another script in a browser that > reads periodically from a file written by the agent. But this seems like > an awkward solution.
Why not just have the agent print to STDOUT or STDERR? > I have very little experience with interprocess control, but I thought I > should be able to launch the spider script to run in the background with > a system call and then redirect to a perl script that will refresh > periodically and display the books (or number of books) retrieved in a > browser window. But I haven't been get the browser to do anything until > the child process is finished. That may be a feature of the browser. Try outputting a lot of blank lines to get it to display. Have you set autoflush on and/or disabled output buffering? > There is a script published in Linux magazine by Randal Schwartz > http://www.linux-mag.com/content/view/1122/2222/ that works along these > lines, but it doesn't work on my Win32 machine. (Its example of a long > process is "exec traceroute" -- the only change I made was to make this > Win32's "exec tracert" ) > > I've tried to find the solution in Learning Perl, the Perl Cookbook and > Perl FAQ, but the material dealing with IPC seems geared to running from > the command line. > > Is there a basic tutorial that covers forking thing somewhere? > > Thx, > > Bruce > > -- > Bruce McKenzie > http://www.2MinuteExplainer.com >