> marks wrote: > > I'm using grep in the install section of my spec file to determine > > if another file does or does not contain a piece of information. > > rpmbuild is aborting with a "bad exit status" message when grep > > does not find the item of interest. > > If you use it directly: > if grep -q bla file; then > .... > > rpmbuild doesn't exit. Of the suggestions this way is my favorite and most clear method. I know that you have the problem solved but I wanted to mention a related problem that I have seen others ask. This is just drift from the original question, but I think related. The return code of grep indicates whether a pattern existed or not and in the case here that is how it was being used. But another common other case is that people are using grep to sift out a particular pattern and the return code is not significant there. Well it is actually significant (disk full problems, permission problems, etc.) but people don't realize it. But the issue I am referring to here is that the 'sh -e' setting will cause a non-zero return from grep to exit the script. grep PATTERN file1 > file2 grep -v PATTERN file1 > file2 This causes a non-zero return code based upon the PATTERN match. > If you want to ignore the exit code of a command, then add ||: (ie. or true). > Note that will change $? to true. I have seen this several times in other people's scripts. I don't like it because it loses I/O errors. grep PATTERN file1 > file2 || true grep -v PATTERN file1 > file2 || true Better to use sed in this case. sed -n /PATTERN/p file1 > file2 sed /PATTERN/d file1 > file2 The return code of sed is related only to errors occuring and not to whether the pattern matched. The sed will fail when there is a real error as desired and the script would terminate in that case. Bob _______________________________________________ Rpm-list mailing list Rpm-list@xxxxxxxxxx https://www.redhat.com/mailman/listinfo/rpm-list
