ebusd/ChangeLog.md

next version: 3.0

Breaking Changes

• switched to C++11 compiler
• separated access level from circuit name
• introduced access control list and user authentication for access to certain messages
• added automatic check for updates (ebusd and configuration)
• use CSV header line for determining columns and picking multi-language columns by language preference
• a circuit name is now required for all message configurations
• raw logging allows logging of messages or each sent/received byte defaulting to messages

Bug Fixes

• corrected numeric condition formatting
• corrected reporting of CSV error position
• corrected last update time of write messages
• close log file on SIGHUP for better logrotate support
• only allow a single scan at the same time
• check allowed value range for base types during decoding
• ignore first seconds of data when calculating symbols per second
• exclude messages without name
• fix for potentially zero resolution
• fix for ebusd not answering scan request from another participant
• corrected missing update notification on master message part
• corrected grouping of JSON output by circuit
• corrected address conflict detection when in answer mode
• fixed potential memory leaks and misused iterator

Features

• added support for MQTT handling via libmosquitto (will be compiled in when library is available)
• allow empty ID in filename for circuit and suffix extraction
• use scan ID as fallback for default circuit name
• added U3N/U3R/S3N/S3R data types
• added another log facility "other"
• added support for using cmake in addition to autoconf
• added Docker support
• allow invalid (non-BCD) coded SW and HW fields when looking for scan config file
• added "auth" command and user/access information to "info" command output
• added user and secret arguments to HTTP port
• use ACL in read/write/find/scan commands
• added -a and -l options to "find" command
• added possibility to define different log level per area on command line and with "log" command
• added decode option to "grab" command presenting decoding hints for unknown messages
• added option to use different source address QQ to read/write/hex commands
• enhanced scan for individual slave
• wait for broadcast scan answers before doing individual scans
• enhanced SymbolString to be aware of master/slave and get rid of CRC (instead calculate while sending/receiving)
• allow unusually formatted MQTT topics
• allow TTQ/TTH types to use less than 8 bits
• added support for optional user-defined columns to config files
• added circuit-level attributes to config files for use in JSON
• added -N option to "read" command

2.4 (2016-12-17)

Bug Fixes

• fixed removed last byte in "hex" command answers
• corrected checking duplicates of chained messages
• fixed reading beyond message bounds
• corrected output precision with reciprocal divisor
• corrected caching of master-master messages
• corrected default dump file name to "/tmp/ebusd_dump.bin"

Features

• extended default definition in CSV to allow message name and comment prefix/suffix
• added automatic reconnect of device if signal loss is persistent and extended "info" result with number of reconnects
• extended "grab" command to allow selection of all or only unknown messages in grab result
• added error log entry when ebusd address conflicts with another participant
• extended BTI/HTI/BTM types to allow replacement value
• added condition name to logging of "!load" instructions and "info" result
• added "--lograwdatafile" and "--lograwdatasize" options logging received/sent bytes to a dedicated text file
• added TTQ base type
• added support for constant value fields

2.3 (2016-11-06)

Bug Fixes

• synchronously request a message if needed for loading "*.inc" files
• corrected derivation for TEM_P type
• allow chained messages to be polled
• removed CRC from slave answer in decode

Features

• no longer poll scan messages since those are expected not to change
• added more message counts to "info" result
• added uih circuit with bar chart containing yields to HTML interface

2.2 (2016-10-16)

Breaking Changes

• completely reworked base data types to allow user-defined types
• added first version of TEM specific data type
• allow inline rename of template reference
• removed "stop" command for enhanced security
• use broadcast ID messages for identification and automatic configuration file selection

Bug Fixes

• corrected extra separator in CSV style dump
• corrected BCD representation in JSON
• corrected serial close()
• fix for storing own prepared master data
• no longer show messages that did not receive data in "find" command
• check address argument for validity in "scan" command
• only use cached data if no input is necessary in "read" command
• fix for poll "continue" errors
• fix for checking command line arguments
• fix for DAY data type
• corrected checking of inline conditions
• escape special characters in string data types for JSON
• corrected dumping of value lists

Features

• allow running multiple ebusd instances by configuration file (/etc/default/ebusd on debian)
• set keepalive option on network device
• added hex option to "find" command
• added optional initial active scan address to "--scanconfig" option that allows initiating a single destination address scan, a full scan, or sending a broadcast ident message (this is now the default)
• added MIN datatype
• avoid output of unprintable characters in string data types
• split verbosity of "read" and "find" commands into 3 levels
• added "indexed" parameter for JSON
• allow non-exact values for truncated time types

2.1 (2016-05-05)

Breaking Changes

• added "hex" command for sending arbitrary hex data, disabled it by default (use "--enablehex" command line option), and limit "write -h" command to known messages only.
  This increases security by avoiding unintended sending of arbitrary messages.
• changed ebusd default address to 0x31 in order to avoid address conflicts with popular devices on the bus
• added "!load" instruction for CSV configuration files allowing conditional loading of single other CSV file
• added "!include" instruction for CSV configuration files in order to include other files
• added support for UDP connected devices

Bug fixes

• require "-c" in write command
• corrected duplicate messages check with condition
• corrected maximum values for BCD and HCD types
• corrected weekday calculation (for smaller systems mainly)
• corrected determination of field name uniqueness
• corrected JSON read of passive messages

Features

• allow CSV to contain empty lines (no field filled in)
• added shorter HCD and BCD data types
• added data type NTS (null terminated string)
• added EXP and EXR data types (exponential)
• added option for changing PID file
• allow using newlines within quoted fields in CSV
• added support for string based conditions and on-the-fly condition definition
• enhanced HTTP user interface
• increased default receive timeout to 25ms
• extract default circuit name from CSV file name
• automatically start grabbing all messages
• add circuit+message name of known messages to grab result
• added buffer for network devices and flush potentially buffered input when connecting
• added option for setting transfer latency and use default 10 ms for network device

2.0 (2016-01-06)

Breaking Changes

• automatic configuration file selection by querying device identification ("--scanconfig").
  Previous configuration files are still usable, but only without the "--scanconfig" command line parameter to ebusd (unless the files are renamed).
  This is now enabled by default in the init.d (or systemd) scripts to work with ebusd-2.x.x configuration files.
• support for chained messages with fields covering more than one message ID

Bug fixes

• avoid multiple identical derived messages
• return "empty" for queries without result (e.g. grab result)
• fix for stopping main loop in rare cases
• fix for extracting default circuit name from CSV file name
• fix for using pselect
• fix for finishing active request when own master address reception timed out
• fix for some memory issues
• fix for closing serial device that is no tty
• fix for dumping message definition
• fix for cache invalidation
• fixes for answer mode

Features

• respond to broadcast "write" command with "done broadcast"
• prepared for full file check with overlapping names and message keys
• added replacement value for date/time types
• extended "info" command with seen addresses, scan state, and loaded CSV file
• improved CSV file check
• insert index suffix extracted from CSV file name when applying defaults (format "ZZ.CCCCC[.index].csv")
• added support for MACH architecture (Mac OS)
• use empty condition messagename for referencing scan message
• allow read and write message definitions having the same ID
• add hint to decoding problems in "read" and "write" if answer was retrieved successfully
• added script for converting "grab result" to CSV
• exclusively lock serial device to prevent simultaneous access by another process
• extended "-i" parameter of "find" command to accept further ID parts
• include decode problems instead of returning invalid JSON
• extended allowed message ID length and max STR and HEX length
• extended "grab" command with "all" option for grabbing all messages instead of only unknown
• allow float values being used for int types
• added big endian data type variants (UIR, SIR, FLR, ULR, SLR)
• use numeric value as fallback for value lists with unknown value association
• increased verbosity during scan
• increased default bus acquisition retries from 2 to 3
• use named objects for JSON fields instead of array with fallback to numbered
• exclude conditions from "find -f" command
• let ebusd answer to scan request (if started with "--answer")
• add special length '*' for consuming remaining input for some string types (IGN, STR, HEX)

Changed files

https://github.com/john30/ebusd/compare/v1.3.0...v2.0

1.3.0 (2015-10-24)

Breaking Changes

• started with a new git repository for working on 2.0 version (previous repository was renamed to ebusd1)
• added support for conditional message definitions based on other message values

Bug fixes

• fix for numeric parsing of empty string
• allow HTTP port query string keys without value
• fix for thread cleanup
• update last message update time when actively sending it
• trim CSV lines and fields
• corrected derivation of field name, comment and unit in templates
• check invalid divisor/values for string and value list based types
• fix for overriding destination address in read command
• fix for cached master data

Features

• new "read" option "-n" for retrieving name/value pairs in numeric form
• new "find" option "-e" for exactly matching name and optional circuit (ignoring case)
• new "find" option "-F" for listing messages in CSV format with selected columns only
• added numeric, verbose, and required options to JSON and use "null" for unset or replacement value
• allow static ".json" files being served by HTTP port
• allow passing raw value when writing name/value pairs
• enhanced static HTML user interface to dynamic
• added possibility to use a different name for referenced field template
• added data type HCD (for Ochsner)
• added base type TTH (truncated time with 30 minutes resolution)
• added base type VTM (reverse HTM)
• added log entry for scan completion
• automatically invalidate cached data of messages with same name+circuit (ignoring level)
• added "info" command
• added circuit and name to verbose "read" output
• added option to pass additional input data to "read" command
• added messages without dedicated destination and cached data decoding error to "find" result
• new "write" option "-d" for overriding destination address
• allow CSV file name to define default destination address and circuit name (format "ZZ.CCCCC.csv") and if present, concat default circuit name and security suffix (circuit name in CSV starting with "#")

Changed files

https://github.com/john30/ebusd/compare/v1.2.0...v1.3.0

1.2.0 (2015-05-25)

Bug fixes

• corrected missing newline in ebusctl
• avoid updating last message change time when only the querying master was modified
• notify running requests on device error

Features

• new command line option "--readonly" for read-only access to the bus
• added support for retrieving data as JSON and additional files via optional HTTP port
• added first version of static HTML user interface (to be extended to a dynamic one later)
• new "read" option "-h" for sending hex read messages
• new "read" option "-p PRIO" for setting the poll priority of a message
• update cache on hex read or write

Changed files

https://github.com/john30/ebusd/compare/v1.1.0...v1.2.0

1.1.0 (2015-03-28)

Bug fixes

• return passive messages returned in "read" regardless of their age
• corrected divisor for chains of derived number types
• corrected master number calculation (no influence on any behaviour)
• fixed signed char warning on some compilers
• corrected minimum lock count according to spec.
• differentiate between valid empty result and not found field name
• fixed scanning for messages in subdirectories for certain file systems (thanks to joltcoke)
• allow mixing references to base types and templates in message definition
• renamed "class" to "circuit" in commands and message definition
• fixed reading cached values
• extended max allowed data length to 25 (spec says 16)

Features

• new command line option "--dumpconfig" for dumping message configurations
• new "find" command option "-f" for retrieving message configurations in CSV format
• new "find" command option "-i PB" to filter messages on primary command byte
• extended "read" command "FIELD" argument with optional ".N" for retrieving the N''th field with that name
• new "write" command option "-c" for compatibility with "read" and "find"
• automatically detect number of masters on the bus and add to "state" command
• support reciprocal divisor
• cache symbols instead of formatted strings. this allows retrieval of individual fields and verbose read from cache
• removed valid CRCs from log and hex "write"
• shortened manufacturer names for "scan"
• allow specifying multiple destination addresses in message definition and defaults
• added Arch Linux PKGBUILD scripts (thanks to cogano)
• new "read" command option "-d ZZ" for overriding destination address (e.g. for manual scan)
• allow message definitions without any field
• added broadcast ident message (requests all masters to send their identification) and added read ident for manual scan
• added support for AUTO-SYN generation (experimental!)
• added "grab" command for collecting and reporting seen unknown messages

Changed files

https://github.com/john30/ebusd/compare/v1.0.0...v1.1.0

1.0.0 (2015-02-18)

Breaking Changes

This is the first version of the completely reworked ebusd since October 2014.

Bug fixes

A lot of problems were fixed, such as memory leaks and segmentation faults.

Features

The daemon is now fully aware of the eBUS protocol and future versions will also support addressing a running daemon as answering master/slave. The main new feature is a completely rewritten configuration file engine, which allows the use of templates and defaults and thus simplifies the configuration files a lot. On the client side, a lot of commands have been added, such as "listen" (automatically send changed values listened to or polled), "find", and "state".

Changed files

https://github.com/john30/ebusd/commits/master

Thanks to Roland Jax and other authors for all their work in previous versions!