sumo¶

What the script does¶

The script has two major purposes:

Manage the dependency database DEPS.DB. It is used to create this file from the output of sumo-scan and to query and change that file.
Create and manage builds. It also keeps note of all builds in the build database BUILDS.DB.

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"
          },
          "make-recipes": {
              "all": [
                  "cd $DIR && ./configure --prefix=.",
                  "$(MAKE) -C $DIR"
              ],
              "clean": [
                  "$(MAKE) -C $DIR realclean"
              ]
          },
          "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>
    },
    "make-recipes": {
        <make-recipes data>
    }
    "releasefile": <releasefilename>,
    "source": {
        <source data>
    },
    "weight": <weight>
}

aliases¶

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,
    ...
]

make-recipes¶

This optional field is used to specify alternative make recipes for the makefile that is generated for all modules of a build.

For each of the make targets “all”, “clean”, “config” and “distclean” a list of lines can be defined that is put in the generated makefile. In the make-recipes map, each of the map keys “all”, “clean”, “config” and “distclean” is optional. For convenience, the string “$DIR” is replaced with the special make variable $(@D) in every line. This is the directory of the checked out module (see also documentation of the “make” command). Note that you do not have to prepend each line with a <TAB> character, sumo already does this.

Note that for the “all” target your last recipe line is usually $(MAKE) -C $DIR.

If you have an empty object (or dictionary in python speak) here, this means that the module has no makefile at all. It is just checked out and possibly configured (see also “commands”).

You have an example of a make-recipes structure at the top of the chapter The dependency database .

You can define make-recipes on the command line with sumo db make-recipes or directly in the dependency database with sumo db edit.

Special variables and characters:

$DIR: (sumo) The directory of the MODULE.
\": (bash) A literal double quote character.
$(VAR): (make) Insert value of make or shell variable VAR.
$$: (make) A dollar character passed to the shell.
\\$$: (make, bash) A literal dollar character passed to the shell.
\\: (json, bash) At the end of the json string this means line continuation for bash.

This is the form of the make-recipes map:

"all": [
    STRING,
    STRING,
    ...
],
"clean": [
    STRING,
    STRING,
    ...
],
"config": [
    STRING,
    STRING,
    ...
],
"distclean": [
    STRING,
    STRING,
    ...
]

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¶

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, COMMANDS means a list of strings that are command lines which are executed in the given order inside the module directory after the module was checked out. Possible patches (see below) are applied after the commands. You may find the feature useful for git sub repositories which must be initialized by an extra git command.

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
.tgz : a tar file compressed with gzip
.tar.bz2 : a tar file compressed with bzip2

The source data has this form:

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

The keys “commands” and “patches” are optional.

“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": {
        "commands": COMMANDS,
        "patches": PATCHFILES,
        "tag": "TAG",
        "url": "REPOSITORY"
    }
}

The keys “commands” and “patches” are optional.

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

The keys “commands” and “patches” are optional.

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

The keys “commands” and “patches” are optional.

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

The keys “commands” and “patches” are optional.

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": {
        "commands": COMMANDS,
        "patches": PATCHFILES,
        "tag": "TAG",
        "url": "REPOSITORY"
    }
}

The keys “commands” and “patches” are optional.

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>

weight¶

This optional field is used to define the weight factor for a module. You usually don’t have to use this, see db weight WEIGHT MODULES for details.

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
incomplete the build is defined but not all module directories are created
broken the build is broken and should be deleted

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 SUBCOMMAND¶

Show the configuration or create or modify a configuration file. These are known subcommands here:

list - list loaded configuration files
show - show configuration data
make - create configuration file
standalone - create configuration for “standalone” builds
local - create configuration for “local” builds
new - create a new configuration file from one of the provided templates

You get help on each subcommand with:

help SUBCOMMAND

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 SUBCOMMAND¶

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.

These are the known subcommands here:

convert: convert a scanfile created by sumo-scan to a DB file
convert-old: convert DB file from old to new format
appconvert: convert a scanfile to a MODULES file for an application
modconvert: print new DB file entries for the given MODULES from a scanfile
edit: edit the dependency file with an editor
format: reformat the dependency file
weight: set the weight factor for modules
alias-add: add an alias for a dependency in a module
dependency-add: add a dependency to a module
dependency-delete: delete a dependency of a module
commands: define commands to be executed after module checkout
make-recipes: define special make-recipes for a module
list: list modules or versions of modules
show: show details of moduleversions
find: search for modules with a regexp
check: consistency check of the DB file
merge: merge two DB files
cloneversion: create a new DB entry by copying an old one
releasefilename: define an alternative filename for the RELEASE file
replaceversion: replace a DB entry with a new one
clonemodule: add a module under a new name in the DB file

You get help on each subcommand with:

help SUBCOMMAND

build SUBCOMMAND¶

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.

These are the known subcommands:

try: check the module specification for completeness and consistency
new: create a new build
remake: do “make clean” and “make all” with a build
find: look for builds that match a module specification
use: use all modules or your module specification in your application
list: list names of all builds
show: show details of a build
state: show or change the state of a build
delete: delete a build

You get help on each subcommand with:

help SUBCOMMAND

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]¶

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 new DIRECTORY TEMPLATE¶

This command is used to create a new sumo directory with a new build directory and a new dependency database.

It creates a new configuration for sumo. DIRECTORY must not yet exist and is created by this command. This command takes all settings and command line options but sets dbdir to DIRECTORY/database. It also sets builddir to DIRECTORY/build. TEMPLATE determines how the dependency databasse file is created. Currently 2 values are known:

empty: Create an empty dependency database.
github: Create a sample dependency database file with some entries for EPICS base, ASYN and the sequencer. In this sample all module repositories are fetched from the internet, mostly github.

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 standalone DIRECTORY¶

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

It creates 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 not given, the command copies the source specification from OLD-VERSION but sets the tag to NEW-VERSION. Note that this is not allowed for “path” and “tar” sources (see below).

If SOURCESPEC is given, the source specification from OLD-VERSION is copied an the all values from SOURCESPEC are added.

A sourcespec has the form::: NAME=VALUE[,VALUE…] [NAME=VALUE[,VALUE..] …]

In general, NAME must start with a letter or underscore character and must be following by a sequence of letters, underscrores or digits.

A VALUE must be a JSON simple value (no map or list). If VALUE is a string, it must be enclosed in double quotes ‘”’ if it contains any of the characters ‘”’, ‘,’ or ‘ ‘.

These are possible names:

type: The source type. Currently known are “path”, “tar”, “cvs”, “svn”, “darcs”, “hg” and “git”.
url: This is the URL. For the types “path” and “tar” it may also be a filename.
tag: This defines the revision tag.
rev: This defines the revision hash key.
patches: This defines names or URLs for patch files. This is the only name, where several values may be provided as a comma separated list.
commands: This defines commands that are executed after the source code is checked out and before any patches are applied.

Note that you can define an empty value (on the bash shell) like in this example:

tag='""'

This means that the “tag” is removed from 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.

db commands MODULE LINES¶

Define commands that are executed after a module is checked out. See also “commands” in the chapter “The dependency database”.

MODULE here is a modulespec of the form MODULE:VERSION that specifies a single version of a module. LINES is a list of space separated strings. It is recommended to enclose the line strings in single or double quotes.

Special variables and characters when you use double quotes:

\": (bash) A literal double quote character.
$(VAR): (make) Insert value of make or shell variable VAR.
$$: (make) A dollar character passed to the shell.
\\$$: (make, bash) A literal dollar character passed to the shell.
\\: (json, bash) At the end of the json string this means line continuation for bash.

db make-recipes MODULE [TARGET] [LINES]¶

Define special make recipes for a module. See also “make-recipes” in the chapter “The dependency database”.

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

If TARGET is given, it must be “all”, “clean”, “config” or “distclean” and specifies the make target for which a recipe is defined. LINES is a list of space separated strings. It is recommended to enclose the line strings in single or double quotes. If LINES is not given, all special rules for the TARGET are removed.

If TARGET (and LINES) are not given, this defines empty make recipes. This has to be done for modules that have no makefile at all. These modules are only checked out by sumo, and possibly configured (see also “commands”).

Special variables and characters when you enclose LINES in double quotes:

$DIR: (sumo) The directory of the MODULE.
\": (bash) A literal double quote character.
$(VAR): (make) Insert value of make or shell variable VAR.
$$: (make) A dollar character passed to the shell.
\\$$: (make, bash) A literal dollar character passed to the shell.
\\: (json, bash) At the end of the json string this means line continuation for bash.

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 clean” 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

There are two ways to activate command completion, described in the following chapters.

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 “,”.

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 have the form ‘REPOTYPE URL’ or ‘type=REPOTYPE url=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 REGEXP, --exclude-states REGEXP`¶

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 NO`¶

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.

`--no-multiprocessing`¶

Do not use multiprocessing in the program. This is mainly here to help debugging the program. Currently multiprocessing is used when the sources for modules of a build are created by checking out from version control systems.

`-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’.

`--editor EDITOR`¶

Specify the preferred editor. If this is not given, sumo takes the name of the editor from environment variables “VISUAL” or EDITOR”.

`--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.

mode	operation	action
get	init	create the repository if it doesn’t exist
	read	none
	write	distr. VCS: commit changes
	write	centr. VCS: none
pull	init	create the repository if it doesn’t exist
	read	distr. VCS: pull
	read	centr. VCS: update
	write	distr. VCS: commit changes
	write	centr. VCS: none
push	init	create the repository if it doesn’t exist
	read	distr. VCS: pull
	read	centr. VCS: update
	write	distr. VCS: pull, commit changes, push
	write	centr. VCS: update, commit changes

sumo¶

What the script does¶

How it works¶

The dependency database¶

aliases¶

dependencies¶

make-recipes¶

releasefile¶

source¶

path¶

tar¶

darcs¶

hg¶

git¶

svn¶

cvs¶

weight¶

The scan database¶

The build database¶

moduledata¶

linkdata¶

state¶

Configuration Files¶

Commands¶

maincommands¶

help COMMAND¶

config SUBCOMMAND¶

lock FILE¶

unlock FILE¶

db SUBCOMMAND¶

build SUBCOMMAND¶

subcommands for maincommand “config”¶

config list¶

config show [OPTIONNAMES]¶

config make FILENAME [OPTIONNAMES]¶

config new DIRECTORY TEMPLATE¶

config standalone DIRECTORY¶

config local DIRECTORY¶

subcommands for maincommand “db”¶

db convert SCANFILE¶

db convert-old OLDDB¶

db appconvert SCANFILE¶

db modconvert SCANFILE MODULES¶

db edit¶

db format¶

db weight WEIGHT MODULES¶

db list MODULES¶

db show MODULES¶

db find REGEXP¶

db check¶

db merge DB¶

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

db releasefilename MODULE RELEASEFILENAME¶

db replaceversion MODULE OLD-VERSION NEW-VERSION¶

db clonemodule OLD-MODULE NEW-MODULE [VERSIONS]¶

db dependency-delete MODULE DEPENDENCY¶

db dependency-add MODULE DEPENDENCY¶

db alias-add MODULE DEPENDENCY ALIAS¶

db commands MODULE LINES¶

db make-recipes MODULE [TARGET] [LINES]¶

subcommands for maincommand “build”¶

build try MODULES¶

build new MODULES¶

build remake BUILDTAG¶

build find MODULES¶

build use MODULES¶

build list¶

build show BUILDTAG¶

build state BUILDTAG [NEW-STATE]¶

build delete BUILDTAGS¶

Command completion¶

Prerequisites¶

Activate command completion on the fly¶

Activate command completion permanently¶

Completion cache files¶

The help pager¶

Options¶

--version¶

-h [OPTIONS], --help [OPTIONS]¶

--summary¶

`--version`¶

`-h [OPTIONS], --help [OPTIONS]`¶

`--summary`¶

`--test`¶

`-c FILE, --config FILE`¶

`-C, --no-default-config`¶

`--disable-loading`¶

`-A, --append OPTIONNAME`¶

`--#preload FILES`¶

`--#opt-preload FILES`¶

`--#postload FILES`¶

`--#opt-postload FILES`¶

`--dbdir DBDIR`¶

`--dbrepomode MODE`¶

`--dbrepo REPOSITORY`¶

`--scandb SCANDB`¶

`--dumpdb`¶

`--logmsg LOGMESSAGE`¶

`-t BUILDTAG, --buildtag BUILDTAG`¶

`--buildtag-stem STEM`¶

`--builddir BUILDDIR`¶

`--localbuilddir BUILDDIR`¶

`-o OUTPUTFILE, --output OUTPUTFILE`¶

`-x EXTRALINE, --extra EXTRALLINE`¶

`-a ALIAS, --alias ALIAS`¶

`-m MODULE, --module MODULE`¶

`-X REGEXP, --exclude-states REGEXP`¶

`-b, --brief`¶

`--recursive`¶

`--detail NO`¶

`-D EXPRESSION, --dir-patch EXPRESSION`¶

`-U EXPRESSION, --url-patch EXPRESSION`¶

`--noignorecase`¶

`--no-checkout`¶

`--no-make`¶

`--makeflags MAKEFLAGS`¶

`--readonly`¶

`--nolock`¶

`--no-multiprocessing`¶

`-p, --progress`¶

`--trace`¶

`--tracemore`¶

`--dump-modules`¶

`--list`¶

`-y, --yes`¶

`--editor EDITOR`¶

`--exceptions`¶

`-v, --verbose`¶

`--version`¶

`-n, --dry-run`¶