Mass rebuild of EPEL6 (2016-01-19)

[Jason L Tibbitts III]

After spending far too much time playing with my mass rebuild script,
and a whole lot of CPU time building packages, I've done another mass
rebuild of EPEL6.  This involves an initial build of EPEL6 (with
testing), another three builds of any failing packages, and then a final
build of the current git of each failed package.  Packages which are
excluded on x86_64 were skipped.

The following packages are skipped because they take absolutely forever
to build:

elpa - Still building after two days on a rather fast machine!
nwchem
pypy

This one's test suite appears to simply hang forever:

rubygem-eventmachine

Missing dependencies:

dinotrace
erlang-erlsom
erlang-ibrowse
freecad
golang-github-endophage-gotuf
golang-github-go-ldap-ldap
golang-github-gonum-blas
golang-github-gonum-floats
golang-github-gorilla-sessions
golang-github-hashicorp-serf
golang-github-lsegal-gucumber
golang-github-prometheus-client_golang
golang-github-spf13-cast
golang-github-spf13-viper
gpredict
jmtpfs
nodejs-ascii-tree
nodejs-duplexify
nodejs-nsp-audit-shrinkwrap
nodejs-nth-check
nodejs-rc
nodejs-registry-url
nodejs-seq
nodejs-silent-npm-registry-client
nodejs-stream-spigot
nodejs-test
ocaml-lwt
perl-Authen-Simple-Passwd
php-doctrine-datafixtures
qt5-qtscript
razorqt
rubygem-fakeweb
runc
tuxcmd



The following fail with "no job control" which indicates an unexpanded
macro (usually) in %build.  All of these appear to be due to the lack of
a %cmake28 macro.

armadillo
biblesync
flexiport
freewrl
fts
hokuyoaist
libccd
liblastfm
nagios-plugins-fts
nss_wrapper
octomap
OpenColorIO
OpenImageIO
pcl
python-gencpp
python-genlisp
python-genmsg
python-genpy
qfaxreader
qpid-proton
rospack
socket_wrapper  
task
trustedqsl
uid_wrapper
urdfdom-headers
vtk
VirtualGL

Test suite (%check) failures:

libgee
libserf
heimdal
nodejs-callsite
nodejs-dateformat
perl-CGI-Untaint
perl-CGI-Untaint-email
perl-Class-MOP
perl-IPC-DirQueue
perl-RPC-XML
perl-RPM2
perl-Test-HTTP-Server-Simple-StashWarnings
perl-Test-MockModule
php-doctrine-orm
php-guzzle-Guzzle
php-pear-Console-Getargs
php-pear-Image-Color
php-pear-Math-Stats
php-pear-Net-URL-Mapper
php-pear-Numbers-Words
php-pear-XML-RSS
php-pear-XML-Serializer
python-tw2-dynforms
python-tw2-excanvas
python-tw2-jqplugins-flot
python-tw2-jqplugins-ui
python-zope-schema
rubygem-icalendar


Plain old compilation failures.  A number of these appear to be
boost-related.

detox
digikam
dmlite-plugins
docker-io
exo
gabedit
gfal
gnome-libs
gnustep-back
gnustep-examples
gnustep-gui
golang-googlecode-tools
gorm
gridengine
ike
ktorrent
lcg-util
libflatarray
libkolabxml
libtranslate
libva-vdpau-driver
libzdb
nqc
ocaml-lablgtk
opendmarc
openoffice.org-diafilter
paraview
perl-PBS
perl-Tk
php-pecl-sphinx
php-twig-Twig
plee-the-bear
portmidi
postgis
publican-fedora
pyactivemq
python-Scriptaculous (missing turbojsonify?)
python-tgcaptcha (missing turbojsonify?)
qt-creator
rubygem-nokogiri
scribus
supertux
tcpextract
tmux-top
tomcat
us-imap
xemacs-packages-extra
xloadimage


Perl packages that try to download dependencies from the net:

perl-Catalyst-Model-DBIC-Schema
perl-Test-WWW-Selenium


Miscellaneous other failures:

ctpl appears to have SCM conflict markers in the spec.
flann fails in %install
mlpack fails to %prep due to a bad patch
ncl references a nonexistent file in %files
plplot appears to reference a nonexistent directory in %files.
quassel fails in %prep (bad patch)


There are some other packages which appear to succeed or fail due to
other random criteria (maybe the number of CPUs the builder has).  I
should be able to find all of these but I've done enough for today.  In
the future I'll just post a list of packages/owners and I guess
eventually file some FTBFS bugs.

 - J<

[Parag Nemade]

Hi,

On Wed, Jan 20, 2016 at 9:46 AM, Jason L Tibbitts III <tibbs(a)math.uh.edu&gt; wrote:
>  After spending far too much time playing with my mass rebuild
> script,
>  and a whole lot of CPU time building packages, I've done another mass
>  rebuild of EPEL6.  This involves an initial build of EPEL6 (with
>  testing), another three builds of any failing packages, and then a final
>  build of the current git of each failed package.  Packages which are
>  excluded on x86_64 were skipped.
>
>  The following packages are skipped because they take absolutely forever
>  to build:
>
>  elpa - Still building after two days on a rather fast machine!
>  nwchem
>  pypy
>
>  This one's test suite appears to simply hang forever:
>
>  rubygem-eventmachine
>
>  Missing dependencies:
>
>  dinotrace
>  erlang-erlsom
>  erlang-ibrowse
>  freecad
>  golang-github-endophage-gotuf
>  golang-github-go-ldap-ldap
>  golang-github-gonum-blas
>  golang-github-gonum-floats
>  golang-github-gorilla-sessions
>  golang-github-hashicorp-serf
>  golang-github-lsegal-gucumber
>  golang-github-prometheus-client_golang
>  golang-github-spf13-cast
>  golang-github-spf13-viper
>  gpredict
>  jmtpfs
>  nodejs-ascii-tree
>  nodejs-duplexify
>  nodejs-nsp-audit-shrinkwrap
>  nodejs-nth-check
>  nodejs-rc
>  nodejs-registry-url
>  nodejs-seq
>  nodejs-silent-npm-registry-client
>  nodejs-stream-spigot
>  nodejs-test 
I checked all above nodejs packages. I have already pushed
(epel-testing) most of them by replacing %{nodejs_arches} with its
actual definition and they are working fine. The only remaining
problematic package is nodejs-nsp-audit-shrinkwrap which I will get
fixed today. I am not sure about nodejs-nth-check which package needs
it.

Regards,
Parag.

[Paul Howarth]

On 20/01/16 04:16, Jason L Tibbitts III wrote:
...
>  Test suite (%check) failures:
>
>  libgee
>  libserf
>  heimdal
>  nodejs-callsite
>  nodejs-dateformat
>  perl-CGI-Untaint
>  perl-CGI-Untaint-email
>  perl-Class-MOP 
I have fixed perl-Class-MOP's test suite and submitted an update for it. 
The issue was caused by an earlier update to perl-Package-Stash, which 
resulted in an error message change that wasn't anticipated by 
perl-Class-MOP's test suite.

Paul.