Bugzilla will be upgraded to version 5.0 on a still to be determined date in the near future. The original upgrade date has been delayed.
Bug 81385 - rpmdeps adds paths to provides of an rpm
rpmdeps adds paths to provides of an rpm
Product: Red Hat Public Beta
Classification: Retired
Component: rpm (Show other bugs)
All Linux
medium Severity medium
: ---
: ---
Assigned To: Jeff Johnson
Mike McLean
Depends On:
  Show dependency treegraph
Reported: 2003-01-08 14:12 EST by ajs
Modified: 2007-04-18 12:49 EDT (History)
0 users

See Also:
Fixed In Version:
Doc Type: Bug Fix
Doc Text:
Story Points: ---
Clone Of:
Last Closed: 2003-01-08 17:29:46 EST
Type: ---
Regression: ---
Mount Type: ---
Documentation: ---
Verified Versions:
Category: ---
oVirt Team: ---
RHEL 7.3 requirements from Atomic Host:
Cloudforms Team: ---

Attachments (Terms of Use)
provides result (12.90 KB, text/plain)
2003-01-08 17:09 EST, ajs
no flags Details
requires result (1.89 KB, text/plain)
2003-01-08 17:11 EST, ajs
no flags Details

  None (edit)
Description ajs 2003-01-08 14:12:31 EST
Description of problem:
rpm-build generates odd 'provides' listing.  Some libraries have relative or
absolute paths in the output of
find /var/tmp/rpm-buildroot -type f | xargs /usr/lib/rpm/rpmdeps --provides. 
This causes all of the libraries provided by the package to not satisfy the
requirements found by rpmdeps --requires.  The rpm thus does not install on the
system on which it was just built.

With rpm-build-4.1-1.06 there were no paths and the rpm would install.

Version-Release number of selected component (if applicable):
rpm -q rpm-build: rpm-build-4.2-0.25.1

I can get rpmbuild to do it with an RPM I build of our local Matlab
installation.  I'm having trouble getting other packages to do it.

Steps to Reproduce:
1. rpmbuild -bi mza-matlab.spec
2. find /var/tmp/mza-matlab-6.5-buildroot -type f | xargs /usr/lib/rpm/rpmdeps
--provides | sort --unique
Actual results:

Expected results:

Additional info:
Comment 1 Jeff Johnson 2003-01-08 17:02:29 EST
I need more details to even begin to guess what's happening.

Can you attach the output of rpmdeps --rpmfcdebug, and point
out the specific file/dependency that is not being generated?
Comment 2 ajs 2003-01-08 17:09:41 EST
Created attachment 89220 [details]
provides result

result of 'find /var/tmp/mza-matlab-6.5-buildroot -type f | xargs
'/usr/lib/rpm/rpmdeps --provides | sort --unique > mza-matlab.provides
Comment 3 ajs 2003-01-08 17:11:23 EST
Created attachment 89221 [details]
requires result

result of 'find /var/tmp/mza-matlab-6.5-buildroot -type f | xargs
/usr/lib/rpm/rpmdeps --requires | sort --unique > mza-matlab.requires'
Comment 4 ajs 2003-01-08 17:18:47 EST
Notice that the provides script says that the rpm provides e.g.
'../../bin/glnx86/libmex.so(libmex.INTERNAL)' while the requires script wants
'libmex.so' and 'libmex.so(libmex.INTERNAL)'.  All of the lines in the provides
result that have paths in them generate complaints when I try and install the rpm.

I'm in the process of getting the output of 'rpmdeps --rpmfcdebug'
Comment 5 ajs 2003-01-08 17:27:15 EST
the output of 'find /var/tmp/mza-matlab-6.5-buildroot -type f | xargs
/usr/lib/rpm/rpmdeps --rpmfcdebug > mza-matlab.rpmfcdebug 2>&1' is about 3 1/2
Meg.  Do you still want it as an attachment?
Comment 6 Jeff Johnson 2003-01-08 17:29:46 EST
Ick. I can see lots of packaging problems here.

Add to spec file (or to build system configuration)
    %_use_internal_dependency_generator 0
    %__find_requires /usr/lib/rpm/find-requires
    %__find_provides /usr/lib/rpm/find-provides

and you'll be autogenerating dependencies
the Good Old Way.

You will lose per-file dependencies, and file
classes/colors, but I very much doubt that you care.
Comment 7 Jeff Johnson 2003-01-08 17:31:46 EST
The --rmfcdebug indicates which files generated
which dependencies.

The next step is to examine readelf -a
output, looking for anomalies.
Comment 8 Jeff Johnson 2003-01-08 17:34:04 EST
Note if adding to spec file you want
    %define __find_requires /usr/lib/rpm/find-requires

Note You need to log in before you can comment on or make changes to this bug.