On Wed, 13 Mar 2002, Tom Robinson wrote: > Should I be searching the 70,000 lines of output for an error or not > worry about it?
Depending on what went wrong, you can often find the culprit within the last dozen lines or so. Otherwise, if you hit some kind of cascading error, then it's usually not too hard to scan-scroll up the listing, since you'll be seeing (generally) some kind of repeating pattern and you want to find the last normal thing before that pattern kicked in. One thing I've been doing lately is using the 'tee' command to save a copy of the build process. As normally used, tee takes its standard input, branches a copy of that to a file, and allows the output to continue on it's output without being altered in any way. As a result, it's a good tool for logging batch processes like this -- just snap it onto the end of you're pipeline: % sudo fink install pkg | tee ~/log/fink_pkg If the build works, delete the package; if it fails then you have a handy place to start your autopsy from. The one snag is that, as used here, it only grabs the previous command's standard output -- not standard error. And of course, if something goes wrong then you want to see that stream. Since I haven't been able to figure out the syntax for redirecting stderr to stdout in tcsh, I've been doing this with bash: $ sudo fink install pkg 2>&1 | tee ~/log/fink_pkg I think you can also log the error separately, but I haven't got the syntax pinned down yet. Something like this though: $ sudo fink install pkg |2>~/log/fink_pkg_err | tee ~/log/fink_pkg Not that this necessarily answers your question, but I'm finding that sifting through potentially very long debug output is a lot easier in a regular text editor than it is by scrolling up & down in a terminal. -- Chris Devers [EMAIL PROTECTED] Apache / mod_perl / http://homepage.mac.com/chdevers/resume/ _______________________________________________ Fink-beginners mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/fink-beginners
