The new format is:
<name>-<source release>-<target release>-<distribution id>-<architecture>.delta.pisi
e.g. pisi-179-180-p11-x86_64.pisi
Release numbers are used instead of build numbers.
The new format is:
<name>-<version>-<release>-<distribution id>-<architecture>.pisi
e.g. pisi-2.3.2-180-p11-x86_64.pisi
We will not use build numbers any more. This commit removes some of the code
related to build numbers. New pisi will not deal with build numbers.
Before the stable release, all buildno-related code will be removed.
It is now possible to use digits after a dash character (e.g. polkit-qt-1).
Distribution id is a short string to represent the distribution release
(e.g. p11 for Pardus 2011).
Architecture strings cannot start with a digit. This is needed to
distinguish old and new packages.
With this option, the compression type of index file can be selected.
It is possible to pass multiple types separated by comma.
e.g. pisi index --skip-signing --compression-types=bz2,xz
If defined, use WorkDir in actions.py. If it is not defined or the
directory does not exist, first try "srcname-version" then the
basename of archive URL after splitting extensions.
With this feature, it is not necessary to hardcode versions/releases
used in inter-dependencies.
<RuntimeDependencies>
<Dependency release="current">xorg-server-common</Dependency>
</RuntimeDependencies>
The "current" value will be replaced by the release of the source
package that this spec file defines. This is done when building
the package so that binary packages and index files do not include
"current" string.
This commit also changes the behavior when a sub-package does not
contain any files. Such packages will not be built and a warning
will be printed for each empty package.
If the key "patterns" do not have any value (null string), than it should be
replaced by "-". But the former condition was not able to catch up those null
strings. The condition has changed to work properly.
The old language function was not easy to read and understand. Splitting it into two
functions makes it more readable. The code itself is rewritten from scratch. The
function's parsing ability was by default faulty which has caused to create error .dat
files.
Texlive 2009 has some significant changes in the source tree.
-The format.cnf file is created from the information of tlpobj files. This is parsed via
the function addFormat().
-buildFormatFiles() has some enviroment changes.
-The files of the source itself is distributed in the source dir. This must be relocated
and re-distributed properly. This relocation information is available in the the tlpobj files.
The function which reads the information and relocate the files is named as moveSource().
-handleConfigFiles() is rewritten to provide simplicity and readability. Also there is no need of
parameters, because the prefixes are the same everywhere.
-The order of the functions has been changed. First config files has to be created, after that it's
possible to build formats