[Bug 187610] Review Request: crm114 - CRM114 Bayesian Spam Detector

Wed Aug 16 02:05:42 UTC 2006

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

Summary: Review Request: crm114 - CRM114 Bayesian Spam Detector

https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=187610

jima at beer.tclug.org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
          Component|Package Review              |osgcal

------- Additional Comments From jima at beer.tclug.org  2006-08-15 21:56 EST -------
Going down my own review checklist:
http://beer.tclug.org/fedora-extras/review-checklist.txt

1. One rpmlint error; will cover below.
2. I have some concerns about the version/release of this package. Will discuss
below.
3. Spec is named crm114.spec, check.
4. Package meets Packaging Guidelines, AFAICT.
5. Licensed under GPL, check.
6. License: GPL, check.
7. n/a, no LICENSE/COPYING file in tarball.
8. Spec appears to be American English.
9. Spec seems legible.
10. md5sum on crm114-20060704a-BlameRobert.no-TRE.src.tar.gz does NOT match
upstream. WTF?
11. Compiles and builds on i386/ppc (my two supported build platforms).
12. x86_64 excluded, as per dependency on tre-devel.  You should know the drill.
13. Builds under Plague, so I imagine all of its dependencies are listed.
14. n/a, I think.
15. n/a (no shared libs)
16. n/a
17. crm114-emacs does not own %{_datadir}/emacs/site-lisp/; I believe emacs-el
does.  Please add a dependency.
18. No duplicate %files entries.
19. Not 100% certain on the defattr for the main package.
20. Has valid %clean section.
21. Macro use appears consistent.
22. Package contains code, not content.
23. The doc directory makes up more than half the package's size; you may want
to consider splitting it off to a -doc subpackage. (As discussed on IRC.)
24. I don't see anything in %doc affecting runtime.
25. No header files or static libraries.
26. No .pc files.
27. No library files, much less ones with suffixes.
28. n/a (no -devel subpackage)
29. No .la files.
30. No GUI applications.
31. Doesn't own any directories owned by other packages (to the best of my
knowledge).
32. Packager should poke upstream to include a LICENSE file.
33. I'm not sure there are any description/summary translations available.
34. Package builds as i386 and ppc in Plague (and thus Mock).
35. Package won't build on x86_64 due to dependency's ExcludeArch: x86_64; other
architectures, yes.
36. I can't verify full functionality, but the binary doesn't segfault on i386/ppc.
37. No scriptlets.
38. The -emacs subpackage doesn't Requires: %{name} = %{version}-%{release};
submitter may want to consider doing so.

 So the standing issues, as I see them (some of which we discussed), are:

- rpmlint returns:
W: crm114-emacs no-documentation
 Not sure it's worth remedying.  I welcome more experienced reviewers to chime
in on this.
- I'm concerned about the versioning scheme.  How will you be able to re-release
20060704a if there's a bug?  Incrementing Release to 2.20060704a will preclude
resetting the first digit to 1 with the next release.  Also, I believe the 'a'
violates the Naming Guidelines.  Again, I welcome outside feedback on this.
- #10 concerns me greatly.  Did you repack the tarball or something?
- I think you're missing ExcludeArch: x86_64 (I wouldn't notice, as I don't have
an x86_64 buildsys, but this has been a subject of discussion).
- Maybe add Req: emacs-el for -emacs, due to that package owning the directory
the file is in.  (I believe you did this already, offline.)
- Is %defattr(-,root,root,-) well-defined enough?
- You might want to split out the documentation to a subpackage.
- Maybe ask upstream to provide a LICENSE file.
- Maybe add Req in #38.

 I'm sure we'll be in touch.

-- 
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.