[gentoo-science] [ANN] g-octave 0.4

[gentoo-science] [ANN] g-octave 0.4

[Rafael Martins]

Hi folks,

I have just released g-octave 0.4, that is probably the last version
released as part of the SoC. Here are the news of this version:

= Py3k support =
g-octave modules and the main script were ported to Py3k and will work
with Python >= 2.6 out of the box. The ebuild will install g-octave
for all the Python versions available in your system.

= Restrict the use of g-octave to the users/groups allowed to build packages =
Now only the users that belongs to the user group of the package
manager (e.g. portage, paludisbuild) are allowed to run g-octave.

= Search of Packages =
A command-line option (-s/--search) was added to the main script. This
option will receive an argument, that is the search term or a regular
expression, and will output a list of matches.

= Package database moved to Git =
The g-octave package database now lives on a Git repository at GitHub
[1] and is managed using a new script, that do almost all the work
automatically. The g-octave main script doesn't depends on Git though,
it uses the snapshot tarballs generated by Github to install the
package database to the user.

= New management of live ebuilds =
The old versions of g-octave did the masking of live ebuilds
implicitly. The user just needed to type 'g-octave packagename-9999'
to install the live version of 'packagename'.. This was changed and
now the user will need to change the value of the variable "use_scm"
to "true" in the configuration file "/etc/g-octave.cfg". But this will
not yet install the live version by default, it will create an ebuild
without keywords. The user will still need to unmask the live version
in the package manager configuration files. I also added 2
command-line options (--scm and --no-scm) that are useful to change
the behavior of g-octave on the command line interface, without touch
the configuration file.

= CLI option '--list' rewritten =
The command-line option '--list' now groups the package versions and
show the data in a better shape for the user. I also added a
--list-raw option, that print all the packages available in the
database to the stdout, one package per line and exit, useful for scripts.

= Added the option '--oneshot' =
I added the command-line option '--oneshot' to g-octave, that have the
same behavior in emerge. I just forgot to add it previously. :)

= New script for automated tests =
I installed a XML-RPC plugin at my trac instance [2] and rewrote my
script to do automated tests, to use XML-RPC instead of parse the page
HTML when reporting build errors.

= Proper management of licenses =
The LICENSE variable of the ebuilds is now filled correctly, based on
the licenses from ${PORTDIR}/licenses

= Added metadata.xml =
g-octave now writes a metadata.xml file on the directory with the
ebuilds, if none.

= New trac instance and domain =
my g-octave stuff just got new home (thanks ferringb) and a cool
domain name ( http://g-octave.org ).

That's it!

The ebuild is available on the science overlay and the updated
documentation is available here:
http://doc.g-octave.org/0.4/

[1] - http://github.com/rafaelmartins/g-octave-db
[2] - http://www.g-octave.org/trac

Sorry for the big read ;-)

Best Regards.

-- 
Rafael Goncalves Martins
http://rafaelmartins.eng.br/