[Rd] bug in install-info.pl ?

Kurt Hornik hornik at ci.tuwien.ac.at
Fri Apr 11 11:37:21 MEST 2003

>>>>> mikalzet  writes:

> I have run across a minor problem which would seem to be a bug in the
> install-info.pl script. My data regard R 1.6.2 but I don't suppose the
> script has been changed so perhaps this could be fixed for R 1.7.0.

> Someone mailed me saying they could not recompile my R 1.6.2 src.rpm for
> mandrake 9.1. This was the error message:

> .....
> Checking for unpackaged file(s): /usr/lib/rpm/check-files /var/tmp/R-root
> error: Installed (but unpackaged) file(s) found:
> ,b  (B /usr/share/info/dir.old.bz2

> RPM build errors:
> ,b   (B Installed (but unpackaged) file(s) found:
> ,b  (B /usr/share/info/dir.old.bz2

> I therefore obtained a version of 9.1 and reproduced the problem.

> Build functions correctly; the packaging process fails. RPM 4.0.4
> provided my mandrake 9.1 contains a script called check-files which
> aborts packaging on discovering, in the directory tree containing the
> files to be packaged, a file which will not be included in the final
> rpm.

> The file in this case is:

> ../R-root/usr/share/info/dir.old.bz2

> Deleting this file after the build process and before the check-files
> script enters into play allows the packaging process to terminate
> correctly.

> This could be an error in the spec file, I suppose, but my impression
> is that this file is created (but not deleted, as it should be) by the
> install-info.pl script. On earlier versions of mandrake the problem
> does not come up because no check-files script exists, and so presence
> of unused files is not noticed.

> Could a perl guru take a look at the script and tell me if my
> diagnosis is correct ?

I don't see any 'bz2' in the install-info.pl script.

Now the configure mechanism first checks for an install-info *program*
and if this is not found (and Perl 5.004 or better was found) uses the
install-info script: can you look at your config.status what
INSTALL_INFO ended up being?


More information about the R-devel mailing list