Bug 9508
| Summary: | %files -f breaks when using RPM shell variables in %setup -n | ||
|---|---|---|---|
| Product: | [Retired] Red Hat Linux | Reporter: | David D. Kilzer <ddkilzer> |
| Component: | rpm | Assignee: | Jeff Johnson <jbj> |
| Status: | CLOSED WORKSFORME | QA Contact: | |
| Severity: | medium | Docs Contact: | |
| Priority: | medium | ||
| Version: | 5.2 | ||
| Target Milestone: | --- | ||
| Target Release: | --- | ||
| Hardware: | i386 | ||
| OS: | Linux | ||
| Whiteboard: | |||
| Fixed In Version: | Doc Type: | Bug Fix | |
| Doc Text: | Story Points: | --- | |
| Clone Of: | Environment: | ||
| Last Closed: | 2000-02-22 10:56:04 UTC | Type: | --- |
| Regression: | --- | Mount Type: | --- |
| Documentation: | --- | CRM: | |
| Verified Versions: | Category: | --- | |
| oVirt Team: | --- | RHEL 7.3 requirements from Atomic Host: | |
| Cloudforms Team: | --- | Target Upstream Version: | |
| Embargoed: | |||
%define FILE_LIST rpmfiles.mft
Name: setup-files-bug
Summary: Demonstrates %setup -n/%files -f bug.
Version: 1.0
Release: 1
Source0: foo.txt
Group: Regression/Test
BuildRoot: /var/tmp/%{name}-buildroot
Copyright: GPL
Prefix: /usr
%description
Demonstrates %setup -n/%files -f bug.
%prep
%setup -c -n $RPM_PACKAGE_NAME -T
%build
%install
mkdir -p $RPM_BUILD_ROOT
cp -p %{SOURCE0} $RPM_BUILD_ROOT/
cd $RPM_BUILD_ROOT
find . -type f | sed -e 's,^\.,\%attr(-\,root\,root) ,' \
> $RPM_BUILD_DIR/$RPM_PACKAGE_NAME/%{FILE_LIST}
%clean
rm -rf $RPM_BUILD_ROOT
%files -f %{FILE_LIST}
%changelog
* Wed Feb 16 2000 David Kilzer <ddkilzer>
- Created spec file
See also comment from oliver.nz on 02/07/00 22:01 in bug 9091. http://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=9091 The problem is in
%setup -c -n $RPM_PACKAGE_NAME -T
The macro %setup does not run in a shell context, so the name is taken
literally as '$RPM_PACKAGE_NAME'.
Use %{name} instead, as that is the macro from which RPM_PACKAGE_NAME
is set.
Actually, %setup works fine (probably by accident) with -n $RPM_PACKAGE_NAME since the literal name is used later in a shell context. It is %files that gets tripped up later in the build process by doing this. It would be nice if this 'feature' were documented since it was a pain to track down, and it is not at all obvious to the casual SPEC file writer. "Works fine" only if you like literally '$RPM_PACKAGE_NAME' as a file name. A spec file is not a shell script, even though sections like %prep of the spec file are shell scriptlets written in shell. Commit pushed to master at https://github.com/openshift/origin https://github.com/openshift/origin/commit/222912269321e0a506452df54c37ac9664fc947f Merge pull request #9583 from rhcarvalho/issue9508-new-app-env Merged by openshift-bot |
If a directory name with an RPM shell variable is used in "%setup -n" with "%files -f", then RPM won't find the file list. For example, if this %setup line is used: %setup -c -n $RPM_PACKAGE_NAME -T with this %files line (and macro definition): %define FILE_LIST rpmfiles.mft %files -f %{FILE_LIST} Then RPM will fail to find the %{FILE_LIST} file. If "$RPM_PACKAGE_NAME" is hard-coded to the actual package name in %setup, then RPM will function correctly. I will attach a SPEC file that may be used as a test case. Note that /usr/src/redhat/SOURCES/foo.txt will need to be touched before running it. I am using rpm-3.0.2-5.x (the rpm-3.x upgrade for Red Hat 5.x users) on Red Hat 5.2.