[Bug 226366] Merge Review: regexp

[bugzilla at redhat.com]

Please do not reply directly to this email. All additional
comments should be made in the comments box of this bug report.

Summary: Merge Review: regexp


https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=226366





------- Additional Comments From vivekl at redhat.com  2007-02-09 14:03 EST -------
(In reply to comment #2)
> MUST:
> X rpmlint on regexp srpm gives no output
> W: regexp non-standard-group Development/Libraries/Java
> Perhaps:  System Environment/Libraries ?
It seems use of the existing group is acceptable:
https://www.redhat.com/archives/fedora-packaging/2007-February/msg00070.html
> X package meets packaging guidelines.
> . BuildRoot incorrect.  As per this:
> %{_tmppath}/%{name}-%{version}-%{release}-root-%(%{__id_u} -n)
Amended.
> . do we need section free?
Its a redundant JPackage artifact, removed.
> X specfile is legible
> . do we still need the crazy gcj_support line?
AFAICR the incantation was added so native compilation (i.e. arch dependence)
could be specified on a build machine directly without the need to modify spec
files. However, brew prevents the use of machine specific settings, hence the
use of the %define at the top. However, if the packages are built on mock, such
settings can be provided on the build machine and the hardcoded %define can be
removed. 
> X source files match upstream
> . I can't find the tarball.  Also, Source0 can be the actual URL ending with the
> tar.gz.
Really? With
Source0:http://www.apache.org/dist/jakarta/regexp/jakarta-regexp-%{version}.tar.gz
wget http://www.apache.org/dist/jakarta/regexp/jakarta-regexp-1.4.tar.gz  brings
in the tar ball fine for me. Note the replacement of %{version} in the URL.
Surely the use of the macro is not a problem?

> X BuildRequires are proper
> . why is jpackage-utils in Requires(pre,post)?
According to the guidelines, all directories created by the package must be
owned by the package or the package must require a package that provides the
directory. Directories like %{_javadir} and %{_javadocdir} (/usr/share/java,
/usr/share/javadoc) are provided by jpackage-utils and since the package tries
to install/uninstall things to these directories, I think the presence of these
directories ought to be mandated for the package to be installed/uninstalled.

> X package owns all directories and files
> . why is the javadoc symlink not just made in %install and then added to the
>   %file section?
Fixed. The %pre and %post scriptlets for the javadoc are there for multiple
versions of the javadoc package to coexist and the unversioned symlink allows
crosslinking of javadocs.

> X final provides and requires are sane
> Do we need a 'java' dependency somewhere?  Does the (erroneous, I think)
> Requires(pre,post) on jpackage-utils imply a regular Requires on it?  Do we
> need things in coreutils (rpm, ln) in Requires(post,postun)?
Added the Requires on java
The Requires(x) on jpackage-utils has been commented on above. As far as the
question of /bin/rm and /bin/ln in the requires(x) is concerned, this is to
ensure that rpm transactions ensure these are present before the
installation/uninstallation of the package since the %pre and %postun scripts
use them.


-- 
Configure bugmail: https://bugzilla.redhat.com/bugzilla/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the QA contact for the bug, or are watching the QA contact.