sumo

What the script does

The script has two major purposes:

The script takes one or mode commands and has a number of options. Single character options always start with a single dash “-”, long options start with a double dash “–”, commands are simple words on the command line.

How it works

Information on all known modules and module versions is kept in the dependency database. This file also contains a source specification for each module that may be a directory, tar file or url or a repository specification.

A complete and consistent set of modules that is compiled is called a build. All builds are kept in a single directory, the support directory. Information on builds is kept in a JSON file, the build database BUILDS.DB.

The dependency database

The dependency database DEPS.DB file is a JSON file that contains information on versions of support modules and their dependencies. Here is an example how this file looks like:

{
  "BSPDEP_TIMER": {
      "R6-2": {
          "aliases": {
              "BASE": "EPICS_BASE"
          },
          "dependencies": [
              "BASE"
          ],
          "source": {
              "darcs": {
                  "tag": "R6-2",
                  "url": "rcsadm@aragon.acc.bessy.de:/opt/repositories/controls/darcs/epics/support/bspDep/timer"
              }
          }
      }
  },
  "MCAN": {
      "R2-4-0": {
          "aliases": {
              "BASE": "EPICS_BASE",
              "MISC_DBC": "DBC",
              "MISC_DEBUGMSG": "DEBUGMSG",
              "SOFT_DEVHWCLIENT": "DEVHWCLIENT"
          },
          "dependencies": [
              "ALARM",
              "BASE",
              "MISC_DBC",
              "MISC_DEBUGMSG",
              "SOFT_DEVHWCLIENT"
          ],
          "source": {
              "darcs": {
                  "tag": "R2-4-0",
                  "url": "rcsadm@aragon.acc.bessy.de:/opt/Epics/R3.14.12/support/mcan/2-4-0"
              }
          }
      },
      "R2-4-1": {
          "aliases": {
              "BASE": "EPICS_BASE",
              "MISC_DBC": "DBC",
              "MISC_DEBUGMSG": "DEBUGMSG",
              "SOFT_DEVHWCLIENT": "DEVHWCLIENT"
          },
          "dependencies": [
              "ALARM",
              "BASE",
              "MISC_DBC",
              "MISC_DEBUGMSG",
              "SOFT_DEVHWCLIENT"
          ],
          "source": {
              "darcs": {
                  "tag": "R2-4-1",
                  "url": "rcsadm@aragon.acc.bessy.de:/opt/Epics/R3.14.12/support/mcan/2-4-0"
              }
          }
      },
  },
}

The basic datastructure is this:

{
    MODULENAME : {
        VERSIONNAME : {
            <versiondata>
        },
        VERSIONNAME : {
            <versiondata>
        },
        ...
    }
}

The versiondata map has this form:

{
    "aliases": {
        <aliasdata>
    },
    "dependencies": {
        <dependency data>
    },
    "releasefile": <releasefilename>
    "source": {
        <source data>
    }
}

aliasdata

When the support module is to be compiled “sumo build” creates a RELEASE file from the known dependencies of the module. The RELEASE file contains variable definitions, one for each dependency whose name is the module name and whose value is the path of the compiled module. If a module needs a variable name that is different from the module name, an alias must be defined. For each dependency that is part of the alias map, the ALIASNAME of the alias map is taken. The aliasdata map has this form:

{
    MODULENAME: MODULEALIAS,
    MODULENAME: MODULEALIAS,
    ...
}

dependencies

This is a list of modules this module depends on. Note that we do not store the versions of the modules here. Information on which version is compatible with another version can be found in the build database BUILDS.DB. This is the form of the dependencies list:

[
    MODULENAME,
    MODULENAME,
    ...
]

releasefile

This optional field is used to specify an alternative name for the generated RELEASE file. The default name, if releasefile is not given, is configure/RELEASE.

source data

source data describes where the sources of a module can be found. It is a map with a single key. The key has one of the following values:

  • path: This specifies a directory with the sources. The sources are copied from that location.
  • tar: This specifies a tar file with the sources. The tar file is fetched and extracted.
  • darcs: This specifies a darcs repository.
  • hg: This specifies a mercurial repository.
  • git: This specifies a git repository.
  • svn This specifies a subversion repository.
  • cvs This specifies a cvs repository.

In the following description of source data, FILEURL means a string that is either the path of a file on the local filesystem or an url of a file with this form:

  • http://
  • ftp://
  • ssh://
  • file://

In the following description of source data, PATCHFILES means a list of strings that are names of patchfiles. These are applied to the source with the patch utility after the source is fetched. The strings specifying patchfiles are FILEURLs.

path

This is used to specify a directory that contains all the sources.

For a directory in the local host, the source data has this form:

{
    "path": "PATH"
}

For a directory on a remote host that can be accessed with ssh, the source data has this form:

{
    "path": "USER@HOST:REMOTEPATH"
}
tar

This is used to specify a tar, gzip tar or bzip tar file that contains the sources. The filename must have one of these extensions:

  • .tar : a simple tar file
  • .tar.gz : a tar file compressed with gzip
  • .tar.bz2 : a tar file compressed with bzip2

The source data has this form:

{
    "tar": {
        "patches": PATCHFILES,
        "url": "FILEURL"
    }
}

The key “patches” is optional. If it is given the patches are applied to the source in the given order.

“TARFILE” may be a filename or an URL with one of these forms:

  • http://
  • ftp://
  • ssh://
  • file://
darcs

This is used to specify a source from a darcs repository.

The source data has this form:

{
    "darcs": {
        "patches": PATCHFILES,
        "tag": "TAG",
        "url": "REPOSITORY"
    }
}

The key “patches” is optional. If it is given the patches are applied to the source in the given order.

The key “tag” is also optional, if it is given it specifies the darcs tag that is used to fetch the source.

The key “url” is a darcs repository specification (see manual of darcs for further information).

hg

This is used to specify a source from a mercurial repository.

The source data has this form:

{
    "hg": {
        "patches": PATCHFILES,
        "rev": "REVISION",
        "tag": "TAG",
        "url": "REPOSITORY"
    }
}

The key “patches” is optional. If it is given the patches are applied to the source in the given order.

The key “rev” is optional, if it is given it specifies the mercurial revision that is used to fetch the source. Note that “rev” and “tag” MUST NOT be given both.

The key “tag” is also optional, if it is given it specifies the mercurial tag that is used to fetch the source. Note that “rev” and “tag” MUST NOT be given both.

The key “url” is a mercurial repository specification (see manual of mercurial for further information).

git

This is used to specify a source from a git repository.

The source data has this form:

{
    "git": {
        "patches": PATCHFILES,
        "rev": "REVISION",
        "tag": "TAG",
        "url": "REPOSITORY"
    }
}

The key “patches” is optional. If it is given the patches are applied to the source in the given order.

The key “rev” is optional, if it is given it specifies the git revision that is used to fetch the source. Note that “rev” and “tag” MUST NOT be given both.

The key “tag” is also optional, if it is given it specifies the git tag that is used to fetch the source. Note that “rev” and “tag” MUST NOT be given both.

The key “url” is a git repository specification (see manual of git for further information).

svn

This is used to specify a source from a subversion repository.

The source data has this form:

{
    "svn": {
        "patches": PATCHFILES,
        "rev": "REVISION",
        "tag": "TAG",
        "url": "REPOSITORY"
    }
}

The key “patches” is optional. If it is given the patches are applied to the source in the given order.

The key “rev” is optional, if it is given it specifies the subversion revision that is used to fetch the source. Note that “rev” and “tag” MUST NOT be given both.

The key “tag” is also optional, if it is given it specifies the subversion tag that is used to fetch the source. Note that “rev” and “tag” MUST NOT be given both. If “tag” is given the string “tags” and the tag name are appended to the repository url.

The key “url” is a subversion repository specification (see manual of subversion for further information).

cvs

This is used to specify a source from a cvs repository.

The source data has this form:

{
    "cvs": {
        "patches": PATCHFILES,
        "tag": "TAG",
        "url": "REPOSITORY"
    }
}

The key “patches” is optional. If it is given the patches are applied to the source in the given order.

The key “tag” is also optional, if it is given it specifies the cvs tag that is used to fetch the source.

The key “url” is the cvs repository specification. In the following “<cvsroot>” means the path of your cvs repository and <module> is the directory within “<cvsroot>” where the module is kept. “<user>” and “<host>” are the username and hostname when you contact your cvs repository via ssh. There are three formats you can use here:

Simple path
This has the form <cvsroot>/<module>
Path with “file” prefix
This has the form file://<cvsroot>/<module>
SSH path
This has the form ssh://<user>@<host>:<cvsroot>/<module>

The scan database

When “sumo-scan all” is used to scan an existing support directory it also gathers information on what version of a module depends on what version of another module. In order to keep this information although the dependency database doesn’t contain versions of dependencies, this information is held in a separate file, the scan database or SCANDB.

Here is an example on how this file looks like:

{
    "AGILENT": {
        "R2-3": {
            "BASE": {
                "R3-14-12-2-1": "scanned"
            }
        }
    },
    "AGILENT-SUPPORT": {
        "R0-10": {
            "BASE": {
                "R3-14-12-2-1": "scanned"
            }
        },
        "R0-11": {
            "BASE": {
                "R3-14-12-2-1": "scanned"
            }
        },
        "R0-12": {
            "BASE": {
                "R3-14-12-2-1": "scanned"
            }
        },
        "R0-9-5": {
            "BASE": {
                "R3-14-12-2-1": "scanned"
            }
        }
    },
    "ALARM": {
        "R3-7": {
            "BASE": {
                "R3-14-12-2-1": "scanned"
            },
            "BSPDEP_TIMER": {
                "R6-2": "scanned"
            },
            "MISC_DBC": {
                "R3-0": "scanned"
            }
        },
        "R3-8": {
            "BASE": {
                "R3-14-12-2-1": "scanned"
            },
            "BSPDEP_TIMER": {
                "R6-2": "scanned"
            },
            "MISC_DBC": {
                "R3-0": "scanned"
            }
        }
    }
}

The basic datastructure is this::

{
    MODULENAME: {
        DEPENDENCY_MODULENAME: {
            DEPENDENCY_VERSION: STATE
            DEPENDENCY_VERSION: STATE
            ...
        }
    }
}

For each dependency of a module this structure contains the version of the dependency and a state. The state can be “stable” or “testing” or “scanned” but is always “scanned” if the file was generated with sumo db.

The build database

The build database BUILDS.DB file is a JSON file that contains information of all builds in the support directory.

Here is an example how this file looks like:

{
    "001": {
        "modules": {
            "ALARM": "R3-5",
            "ASYN": "R4-15-bessy2",
            "BASE": "R3-14-8-2-0",
            "BSPDEP_CPUBOARDINIT": "R4-0",
            "BSPDEP_TIMER": "R5-1",
            "CSM": "R3-8",
            "EK": "R2-1",
            "GENSUB": "PATH-1-6-1",
            "MCAN": "R2-3-18",
            "MISC": "R2-4",
            "SEQ": "R2-0-12-1",
            "SOFT": "R2-5",
            "VXSTATS": "R2-0"
        },
        "state": "stable"
    },
    "002": {
        "linked": {
            "ASYN": "001",
            "BASE": "001",
            "BSPDEP_CPUBOARDINIT": "001",
            "BSPDEP_TIMER": "001",
            "CSM": "001",
            "EK": "001",
            "GENSUB": "001",
            "MISC": "001",
            "SEQ": "001",
            "SOFT": "001",
            "VXSTATS": "001"
        },
        "modules": {
            "ALARM": "R3-4",
            "ASYN": "R4-15-bessy2",
            "BASE": "R3-14-8-2-0",
            "BSPDEP_CPUBOARDINIT": "R4-0",
            "BSPDEP_TIMER": "R5-1",
            "CSM": "R3-8",
            "EK": "R2-1",
            "GENSUB": "PATH-1-6-1",
            "MCAN": "R2-3-18",
            "MISC": "R2-4",
            "SEQ": "R2-0-12-1",
            "SOFT": "R2-5",
            "VXSTATS": "R2-0"
        },
        "state": "unstable"
    }
}

The basic datastructure is this:

{
    BUILDTAG : {
        <builddata>
        },
    BUILDTAG : {
        <builddata>
        },
    ...
}

The builddata has this form:

{
    "linked": {
        <linkdata>
        },
    "modules": {
        <moduledata>
        },
    "state": <state>
}

moduledata

moduledata is a map that maps modulenames to versionnames. This specifies all the modules that are part of the build. Since a build may reuse modules from another build not all modules from this map may actually exist as separate directories of the build. The moduledata has this form:

{
    MODULENAME: VERSIONNAME,
    MODULENAME: VERSIONNAME,
    ...
}

linkdata

linkdata is a map that maps modulenames to buildtags. This map contains all modules of the build that are reused from other builds. If a build has no linkdata, the key “linked” in builddata is omitted. The linkdata has this form:

{
    MODULENAME: BUILDTAG,
    MODULENAME: BUILDTAG,
    ...
}

state

This is a state string that describes the state of the build. Here are the meanings of the state string:

  • unstable: the build has been created but not yet compiled
  • testing: the build has been compiled successfully
  • stable: the build has been tested in production successfully
  • disabled the build should no longer be used

Configuration Files

Many options that can be given on the command line can be taken from configuration files. For more details see “configuration files “.

Commands

You always have to provide sumo with a maincommand. Some maincommands need to be followed by a subcommand.

maincommands

help [command...]

This command prints help for the given command. It can be invoked as:

help
help MAINCOMMAND
help SUBCOMMAND
help MAINCOMMAND SUBCOMMAND

You get a list of all known MAINCOMMANDS with:

help maincommand

config

This is the maincommand for all operations for configuration files.

lock FILE

Lock a FILE, then exit sumo. This is useful if you want to read or write a database file without sumo interfering. Don’t forget to remove the lock later with the “unlock” command.

This command must be followed by a filename.

unlock FILE

Unlock a FILE, then exit sumo. If you locked a database with “lock” before you should always unlock it later, otherwise sumo can’t access the file.

This command must be followed by a filename.

db

This is the maincommand for all operations that work with the dependency database or DEPS.DB file.

For all of the db subcommands you have to specify the dependency database directory with option --dbdir or a configuration file.

build

This is the maincommand for all operations that work with builds and the build database (BUILDS.DB).

For all of the build subcommands you have to specify the dependency database directory and the build directory with --dbdir and --builddir or a configuration file.

subcommands for maincommand “config”

config list

List all configuration files that were loaded.

config show [OPTIONNAMES]

Show the configuration in JSON format. OPTIONNAMES is an optional list of long option names. If OPTIONNAMES are specified, only options from this list are saved in the configuration file.

config make FILENAME [OPTIONNAMES]

This command is used to create a new sumo directory with an independent build directory and an independent copy of the dependency database.

Create a new configuration file from the options read from configuration files and options from the command line. If FILENAME is ‘-‘ dump to the console. OPTIONNAMES is an optional list of long option names. If OPTIONNAMES are specified, only options from this list are saved in the configuration file.

config standalone DIRECTORY

Create a new configuration for “standalone” builds. DIRECTORY is created if it does not yet exist. This command takes all settings and command line options but sets dbrepomode to “pull” and dbdir to DIRECTORY/database. It also sets builddir to DIRECTORY/build. Option dbrepo must be set, this is used to create a local copy of the dependency database in DIRECTORY/database. If there is a file “sumo.config” in the current working directory it is copied to “sumo.config.bak”. A new file “sumo.config” is then created in the current working directory.

config local DIRECTORY

This command is used to create a new sumo directory with a new build directory but using existing builds from your current build directory. It also creates an independent copy of the dependency database.

DIRECTORY is created if it does not yet exist. This command takes all settings and command line options but sets dbrepomode to “pull” and dbdir to DIRECTORY/database. It also sets localbuilddir to DIRECTORY/build. Option dbrepo must be set, this is used to create a local copy of the dependency database in DIRECTORY/database. If there is a file “sumo.config” in the current working directory it is copied to “sumo.config.bak”. A new file “sumo.config” is then created in the current working directory.

subcommands for maincommand “db”

db convert SCANFILE

Convert a scanfile that was created by by “sumo-scan all” to a new dependency database. If SCANFILE is a dash “-”, the program expects the scanfile on stdin. Note that options --dbdir and --scandb are mandatory here. With --dbdir you specify the drectory where the new created dependency database file is stored, with --scandb you specify the name of the scan database file. The scan database file contains information on what moduleversion can be used with what dependency version.

db convert-old OLDDB

Convert a dependency database file OLDDB from old to new format. The old format had architecture data (“arch”) for each moduleversion. In the new format this data is removed. Note that this command IGNORES option --dbrepo" it directly operates on the dependency database file in the directory given with option --dbdir.

db appconvert SCANFILE

Convert a scanfile that was created by applying “sumo-scan all” to an application to a list of aliases and modulespecs in JSON format. If SCANFILE is a dash “-” the program expects the scanfile on stdin. The result is printed to the console.

db modconvert SCANFILE MODULES

Convert a scanfile that was created by applying “sumo-scan all” to the dependency database format for all the selected modules. If SCANFILE is a dash “-” the program expects the scanfile on stdin. The result is printed to the console. This data can be added to the dependency database using the command db edit.

db edit

Start the editor specified by option --editor or the environment variables “VISUAL” or “EDITOR” to edit the dependency database file. This command first aquires a file-lock on the file, that prevents other users from acessing the file at the same time. When the editor program is terminated, sumo checks if the file is still a valid JSON file. If not, you can start the editor again or abort the program. If the file is valid JSON, sumo commits the changes if option --dbrepo was specified. If option --logmsg was given, this is used as commit log message, otherwise an editor is started where you can enter a log message. Finally the file lock is released. If you want to edit the dependency database file you should always do it with this command.

db format

Just load and save the dependency database. This ensures that the file is formatted in the standard sumo format. This is useful when the file was edited and you want to ensure that key sort order and indentation are restored. If you specified a repository with --dbrepo, the command will commit the changes. If you want a log message different from “db format” use option --logmsg

db weight WEIGHT MODULES

Set the weight factor for modules. A weight determines where a module is placed in the generated RELEASE file. Modules there are sorted first by weight, then by dependency. Parameter MODULES is a list of modulespecs. Use modulename:{+-}versionname to select more versions of a module.

Note that this command does not use the --modules command line option.

Parameter WEIGHT must be an integer.

db list [MODULES]

If called with no argument, list the names of all modules. If called with ‘.’, the wildcard symbol, list all versions of all modules. If called with argument MODULES, a list of modulespecs MODULE:{+-}VERSION that specifies modules and versions, list all the matching versions of all specified modules.

db show [MODULES]

This command prints only the parts of the dependency database that contain the given modules.

Parameter MODULES is a list of modulespecs MODULE:{+-}VERSION that specifies the modules and versions to operate on.

db find REGEXP

This command shows all modules whose names or sources match a regexp. Parameter REGEXP is a perl compatible regular expression.

db check

Do some consistency checks on the dependency database file in the directory specifed by --dbdir.

db merge DB

Merge the given dependency database file with the dependency database in the directory specifed by --dbdir.

db cloneversion MODULE OLD-VERSION NEW-VERSION [SOURCESPEC]

This command adds a new version of a module to the dependency database by copying the old version. MODULE here is just the name of the module since the version follows as a separate argument. If sourcespec is given, the command changes the source part according to this parameter. A sourcespec has the form “path PATH”, “tar TARFILE [PATCHES]”, “REPOTYPE URL” or “REPOTYPE URL TAG [PATCHES]”. REPOTYPE may be “darcs”, “hg” or “git”. Both, URL or TAG may be ”.”, in this case the original URL or TAG remain unchanged. PATCHES is a list of patchfiles or URLs of patchfiles. If sourcespec is not given, the command adds NEW-VERSION as new tag to the source specification. The command always asks for a confirmation of the action unless option “-y” is used.

db releasefilename MODULE RELEASEFILENAME

This command defines an alternative filename for the RELEASE file of the module. Usually the RELEASE file is generated as “configure/RELEASE”. You can specify a different filename for the given module with this command. This may be useful for support modules that have no regular EPICS makefile system or for some special configurations of the EPICS base. If you set the RELEASEFILENAME to an empty string or “configure/RELEASE”, the special entry for the filename is removed for this module in the dependency database.

db replaceversion MODULE OLD-VERSION NEW-VERSION

This command replaces a version of a module with a new version. MODULE here is just the name of the module since the version follows as a separate argument. All the data of the module is copied. If sourcespec is given, the command changes the source part according to this parameter. A sourcespec has the form “path PATH”, “tar TARFILE”, “REPOTYPE URL” or “REPOTYPE URL TAG”. REPOTYPE may be “darcs”, “hg” or “git”. Both, URL or TAG may be ”.”, in this case the original URL or TAG remains unchanged.

db clonemodule OLD-MODULE NEW-MODULE [VERSIONS]

Copy all versions of the existing old module and add this with the name of thew new module to the dependency database. OLD-MODULE and NEW-MODULE here are just the module names since the versions may follow as a separate argument. If there are no versions specified, the command copies all existing versions. Note that this DOES NOT add the new module as dependency to any other modules.

db dependency-delete MODULE DEPENDENCY

Delete a dependency of a module. MODULE here is a modulespec of the form MODULE:VERSION that specifies a single version of a module.

db dependency-add MODULE DEPENDENCY

Add a dependency to a module. MODULE here is a modulespec of the form MODULE:VERSION that specifies a single version of a module.

db alias-add MODULE DEPENDENCY ALIAS

Define a new alias for a dependency of a module. MODULE here is a modulespec of the form MODULE:VERSION that specifies a single version of a module.

subcommands for maincommand “build”

build try MODULES

This command is intended to help you create module specifications for the “new” command.

Each MODULE here is a modulespec of the form MODULE or MODULE:{+-}VERSION that specifies just a module name, a module and some versions or a single version. You can specify an incomplete list of modules.

The detail of the output is determined by option --detail which is an integer between 0 and 3. 0, the default, gives the shortest, 3 gives the longest report. The program then shows which modules you have to

In any case the command shows which modules are missing since they depend on other modules of your specification and which ones are missing an exact version.

If you converted an existing support directory to sumo you have a scan database file which you can specify with option --scandb to this command.

For a detailed example see try example.

build new MODULES

This command creates a new build. Each module given in MODULES here is a modulespec of the form MODULE:VERSION that specifies a single version of a module. If the buildtag is not given as an option, the program generates a buildtag in the form “AUTO-nnn”. A new build is created according to the modulespecs. Your modulespecifications must be complete and exact meaning that all dependencies are included and all modules are specified with exactly a single version. Use command “try” in order to create module specifications that can be used with command “new”. This command calls “make” and, after successful completion, sets the state of the build to “testing”. If you want to skip this step, use option --no-make. In order to provide arbitrary options to make use option --makeflags.

build remake BUILDTAG

This command recreates a build by first calling “make distclean” and then “make all” with the build’s makefile. If you develop a support module (see also “config standalone” and “config local”) you want to recompile the build after changes in the sources. In order to provide arbitrary options to make use option --makeflags.

build find MODULES

This command is used to find matching builds for a given list of modulespecs. Each module in MODULES here is a modulespec of the form MODULE or MODULE:{+-}VERSION that specifies just a module name, a module and some versions or a single version. The command prints a list of buildtags of matching builds on the console. If option --brief is given, the program just shows the buildtags.

build use MODULES

This command creates a configure/RELEASE file for an application. Each module given in MODULES here is a modulespec of the form MODULE:VERSION that specifies a single version of a module. If option --buildtag is given, it checks if this is compatible with the given modules. Otherwise it looks for all builds that have the modules in the required versions. If more than one matching build found it takes the one with the alphabetically first buildtag. The RELEASE file created includes only the modules that are specified. Output to another file or the console can be specified with option ‘-o’.

build list

This command lists the names of all builds.

build show BUILDTAG

This command shows the data of a build. The buildtag must be given as an argument.

build state BUILDTAG [NEW-STATE]

This command is used to show or change the state of a build. The buildtag must be given as an argument. If there is no new state given, it just shows the current state of the build. Otherwise the state of the build is changed to the given value. If a build is set to state ‘disabled’, all dependend builds are also set to this state. In this case, unless option ‘-y’ or ‘–recursive’ are given, sumo asks for your confirmation.

build delete BUILDTAGS

The directories of the builds are removed and their entries in the build database are deleted. If other builds depend on the builds to be deleted, the command fails unless option ‘–recursive’ is given. In this case all dependent builds are deleted, too. The buildtags must be given as an argument.

Command completion

Command completion means that you can press <TAB> on any incomplete sumo command and you get a list of possibilities how to complete that command. By pressing <TAB> several times you can try each possible completion.

Prerequisites

Command completion works with bash or zsh (Z-Shell), you need to have one of these installed. Your environment variable SHELL must be set to the binary file of the shell, e.g. /usr/bin/bash or /usr/bin/zsh.

In any case the package bash-completion must be installed.

If you use the Z-Shell, the following commands must be executed at start up. Add them for example to the file $HOME/.zshenv:

autoload -U +X compinit && compinit
autoload -U +X bashcompinit && bashcompinit

Setting sumo command completion

There are two ways to do this:

Activate command completion on the fly

Enter this command::
eval `sumo help completion-line`

Activate command completion permanently

Enter this command::
sumo help completion-script > $HOME/_sumo
Then add the line::
source $HOME/_sumo

to your $HOME/.bashrc or $HOME/.zshrc

Completion cache files

Sumo will create cache files in your home directory to speed up command completion. These are the files ”.dbcache.sumo” and ”.buildcache.sumo”. If you don’t want this set the environment variable “SUMOHELP” in a way that it contains the string “nocache” like in:

export SUMOHELP="nocache"

If there are other help options defined in SUMOHELP, you should seperate them with commas ”,”.

The help pager

The build in pager allows you to navigate in long help texts that sumo displays when you use command “help” or option “-h”. There are three modes:

pager:off
The pager is off, all help is printed directly to the console.
pager:on
The pager is used only for long help texts (more than 24 lines).
pager:always
The pager is always used, even for short help texts.

Mode “pager:on” is the default.

You define the pager mode by adding one of the three strings to the environment variable “SUMOHELP” like in:

export SUMOHELP="pager:off"

If there are other help options defined in SUMOHELP, you should seperate them with commas ”,”.

Options

Here is a short overview on command line options:

--version
show program’s version number and exit
-h [OPTIONS], --help [OPTIONS]
If other OPTIONS are given, show help for these options. If OPTIONS is ‘all’, show help for all options. If OPTIONS is missing, show a short generic help message for the program.
--summary
Print a summary of the function of the program.
--test
Perform some self tests.
-c FILE, --config FILE
Load options from the given configuration file. You can specify more than one of these. Unless –no-default-config is given, the program always loads configuration files from several standard directories first before it loads your configuration file. The contents of all configuration files are merged.
-C, --no-default-config
If this option is not given and –no-default-config is not given, the program tries to load the default configuration file sumo-scan.config from several standard locations (see documentation on configuration files).
--disable-loading
If given, disable execution of load commands like ‘#preload’ in configuration files. In this case these keys are treated like ordinary keys.
-A, --append OPTIONNAME
If an option with name OPTIONNAME is given here and it is a list option, the list from the command line is appended to the list from the configuration file. The default is that options from the command line override option values from the configuration file.
--#preload FILES
Specify a an ‘#preload’ directive in the configuration file. This option has only a meaning if a configuration file is created with the ‘makeconfig’ command. ‘#preload’ means that the following file(s) are loaded before the rest of the configuration file.
--#opt-preload FILES
This option does the same as –#preload but the file loading is optional. If they do not exist the program continues without an error.
--#postload FILES
Specify a an ‘#postload’ directive in the configuration file. This option has only a meaning if a configuration file is created with the ‘makeconfig’ command. ‘#postload’ means that the following file(s) are loaded after the rest of the configuration file.
--#opt-postload FILES
This option does the same as –#postload but the file loading is optional. If they do not exist the program continues without an error.
--dbdir DBDIR
Define the directory where the dependency database file ‘DEPS.DB’ is found. A default for this option can be put in a configuration file.
--dbrepomode MODE

Specify how sumo should use the dependency database repository. There are three possible values: ‘get’, ‘pull’ and ‘push’. Mode ‘get’ is the default. The meaning depends on the used version control system (VCS), if it is distributed (git,mercurial,darcs) or centralized (subversion,cvs). There are three possible operations on the dependency database:

  • init : create the dependency database if it doesn’t exist
  • read : read the dependency database
  • write: write (change) the dependency database

Here is what happens during these operations depending on the mode:

mode operation action
get init create the repository if it doesn’t exist
read none
write distr. VCS: commit changes
centr. VCS: none
pull init create the repository if it doesn’t exist
read distr. VCS: pull
centr. VCS: update
write distr. VCS: commit changes
centr. VCS: none
push init create the repository if it doesn’t exist
read distr. VCS: pull
centr. VCS: update
write distr. VCS: pull, commit changes, push
centr. VCS: update, commit changes
--dbrepo REPOSITORY
Define a REPOSITORY for the db file. REPOSITORY must consist of ‘REPOTYPE URL’, REPOTYPE may be ‘darcs’, ‘hg’ or ‘git’. Option –dbdir must specify a directory that will contain the repository for the db file. Before reading the db file a ‘pull’ command will be executed. When the file is changed, a ‘commit’ and a ‘push’ command will be executed. If the repository doesn’t exist the program tries to check out a working copy from the given URL. A default for this option can be put in a configuration file.
--scandb SCANDB
Specify the (optional) SCANDB file. The scan database file contains information on what moduleversion can be used with what dependency version.
--dumpdb
Dump the modified db on the console, currently only for the commands “weight”, “merge”, “cloneversion” and “replaceversion”.
--logmsg LOGMESSAGE
Specify a logmessage for automatic commits when –dbrepo is used.
-t BUILDTAG, --buildtag BUILDTAG
Specify a buildtag.
--buildtag-stem STEM
Specify the stem of a buildtag. This option has only an effect on the commands ‘new’ and ‘try’ if a buildtag is not specified. The program generates a new tag in the form ‘stem-nnn’ where ‘nnn’ is the smallest possible number that ensures that the buildtag is unique.
--builddir BUILDDIR
Specify the support directory. If this option is not given take the current working directory as support directory. A default for this option can be put in a configuration file.
--localbuilddir BUILDDIR
Specify a local support directory. Modules from the directory specifed by –builddir are used but this directory is not modfied. All new builds are created in the local build directory and only the build database file there is modified.
-o OUTPUTFILE, --output OUTPUTFILE
Define the output for command ‘use’. If this option is not given, ‘use’ writes to ‘configure/RELEASE’. If this option is ‘-‘, the command writes to standard-out.”,
-x EXTRALINE, --extra EXTRALLINE
Specify an extra line that is added to the generated RELEASE file. A default for this option can be put in a configuration file.
-a ALIAS, --alias ALIAS
Define an alias for the command ‘use’. An alias must have the form FROM:TO. The path of module named ‘FROM’ is put in the generated RELEASE file as a variable named ‘TO’. You can specify more than one of these by repeating this option or by joining values in a single string separated by spaces. A default for this option can be put in a configuration file.
-m MODULE, --module MODULE
Define a modulespec. If you specify modules with this option you don’t have to put modulespecs after some of the commands. You can specify more than one of these by repeating this option or by joining values in a single string separated by spaces. A default for this option can be put in a configuration file.
-X, --exclude-states
For command ‘try’ exclude all ‘dependents’ whose state does match one of the regular expressions (REGEXP).
-b, --brief
Create a more brief output for some commands.
--recursive
For command ‘build delete’, delete all dependend builds, too. For command ‘build state’ with state ‘disabled’, disable all dependend builds, too.
--detail
Control the output of command ‘try’. The value must be an integer between 0 (very short) and 3 (very long).”
-D EXPRESSION, --dir-patch EXPRESSION
Specify a directory patchexpression. Such an expression consists of a tuple of 2 python strings. The first is the match expression, the second one is the replacement string. The regular expression is applied to every source path generated. You can specify more than one patchexpression. A default for this option can be put in a configuration file.
-U EXPRESSION, --url-patch EXPRESSION
Specify a repository url patchexpression. Such an expression consists of a tuple of 2 python strings. The first is the match expression, the second one is the replacement string. The regular expression is applied to every source url generated. You can specify more than one patchexpression. A default for this option can be put in a configuration file.
--noignorecase
For command ‘find’, do NOT ignore case.
--no-checkout
With this option, “new” does not check out sources of support modules. This option is only here for test purposes.
--no-make
With this option, “new” does not call “make”.j
--makeflags MAKEFLAGS
Specify extra option strings for make You can specify more than one of these by repeating this option or by joining values in a single string separated by spaces. A default for this option can be put in a configuration file.
--readonly
Do not allow modifying the database files or the support directory. A default for this option can be put in a configuration file.
--nolock
Do not use file locking.
-p, --progress
Show progress of some commands on stderr. A default for this option can be put in a configuration file.
--trace
Switch on some trace messages.
--tracemore
Switch on even more trace messages.
--dump-modules
Dump module specs, then stop the program.
--list
Show information for automatic command completion.
-y, --yes
All questions the program may ask are treated as if the user replied ‘yes’.
--exceptions
On fatal errors that raise python exceptions, don’t catch these. This will show a python stacktrace instead of an error message and may be useful for debugging the program.”
-v, --verbose
Show command calls. A default for this option can be put in a configuration file.
--version
Show the program version and exit.
-n, --dry-run
Just show what the program would do.

Table Of Contents

Previous topic

sumo-scan

Next topic

Using sumo in your application

This Page