it would be very useful if linklint would use different exit codes for "everything is fine" vs. "some warnings/errors". regards az
Hi Alexander, IMHO having a non-zero return code when warnings are encountered (as do 'tidy', for instance) is very painful because it breaks 'make' runs; having non-zero return code with errors looks like a good idea. Denis
On Wed, 20 Nov 2002 10:17:47 +0100, Denis Barbier writes: I'm not completely agreeing :-) If an exit code is returned, then you can still ignore it *really* easily - just do '|| true' or use -linklint in a Makefile. As of now the only way to figure out whether everything is ok is to parse the summary. This is wasteful and just not good if you want to run linklint periodically from cron and have an email alert if and when there is a problem. (By problem I mean both errors like dead links as well as warning-stuff.) regards az
Alexander Zangerl once said:
counted in the summary. If we change the exit status as suggested
then might not some working scripts or makefiles break? We could
add a flag to change the exit status. Here are three possibilities:
(A) Just make the change so LinkLint always exits with an error code
if there were any errors or warnings (as counted in summary).
(B) Add (yet another) flag -exit_error (or something). Without this
command line flag LinkLint acts as it does now. With this flag
you get the behaviour as in (A) above.
(C) Add two command line flags with numerical parameters:
-exit_error 0 -exit_warning 3
This would say to give error exit code if there was any error
(error count > 0) or if there were more than 3 warnings. This
way the user can have a quiescent state that has some errors or
warnings that she is choosing to ignore.
Any thoughts?
On Wed, 20 Nov 2002 11:53:48 PST, Jim Bowlin writes: Ah yes, sorry that this was not made clear earlier. Yes, I meant "linklint detects errors on your pages". (I take it as pretty much granted that any program exits with a meaningful indication when its task could not be fulfilled.) Sounds like a good, workable idea; that way backwards-compatibility is maintained, and the functionality I'd like to see is available as well. regards az