GitLab

- coreutils added, we need a unique base64

- core to find tarball of package moved to function and improved

- use the term 'group' instead of 'family'
- prefix 'PSI_' changed to 'PMODULES_' for several shell vars.

- create $PREFIX, make expect the existance of it

- added

- export vars via function 'export_env()'
- PSI_RELEASES renamed to PSI_DEFINED_RELEASES
- bugfixes
- sub-command 'clear' fixed
- sub-command 'whatis' partially fixed

- PSI_RELEASES renamed to PSI_DEFINED_RELEASES
- set PSI_DEFAULT_RELEASES instead of PSI_USED_RELEASES
- PSI_USED_RELEASES renamed to USED_RELEASES (and saved in PMODULES_ENV)

- expose less env.vars to the user, cache them in PMODULES_ENV

- bugfix in sub-command use:
	- PSI_LOADEDFAMILIES was not updated

- bugfix: accept tilde and slash as first character of an argument

- bugfix in subcommand_search: argument --src was not handled correctly

- header of delete_module() fixed
- in function subcommand_sync():
    - use value of PMODULES_INSTALL_SOURCE for src_prefix if not
      specified on command line

- Exit with correct return value in non library mode

Conflicts:
	scripts/Bootstrap/Pmodules/libpmodules.bash

Conflicts:
	scripts/Bootstrap/Pmodules/libpmodules.bash
	scripts/Bootstrap/Pmodules/modmanage.in

- bugfixes
- minor changes

- bugfix in handling sub-command without arguments

- cache 'FAMILES' and 'HIERARCHY_DEPTHS'
- usage output reviewed
- append_path (), prepend_path(), remove_path() moved to libpmodules
- is_modulefile() added (testing shebang)
- argument/option handling reviewed, using 'getopt' now for all sub-commands
- in sub-command 'load': testing whether a module is loadable improved
- use 'info' for most output
- bugfixes

- set 'bindir' if not set
- '\n' removed from info() and error()
- die() always prints to stderr
- get_abspath(), append_path(), prepend_path(), remove_path() added

- exit if a build failed

- bugfixes in setting up family HDF5_serial
- skip call of em.cleanup_build() if nothing was build