Laney

• ⚠️ consume: Renamed the message field to value for consistency with produce and Kafka in general.
• consume, topic #23: Ensured thread safe printing.
• consume, produce, topic: Added the option to print sarama's logout to stderr via -verbose flag.
• consume, produce #24: Enabled reading and producing null values.
• produce: Fixed race condition around reading EOF.
• produce: Now determines the leader for each partition of a given topic and groups the messages per batch into requests to the respective partition's leader.

Krar

• consume #19 #20: the syntax for offsets has changed. The goal is to be more explicit and more flexible. Special values are now denoted by strings, rather than negative values: all to refer to all partitions, newest to refer to the newest available offset of a partition and oldest for the oldest offset respectively. Partitions are now separated by = and start and end offsets by :. For example: all=1:10 to consume messages between offsets 1 and 10 (both sides inclusive) for all partitions. This syntax change allows for relative offset definitions either to skip messages since the oldest offset (e.g. +10), or refer to messages before the newest offset (e.g. -10). Please see kt consume -help for more information. This change is not backwards-compatible.
• consume: The partitions are now consumed in parallel right after they are identified to decrease initial delay.
• topic #20: The names for oldest and newest offset when -partition is specified are now consistent with the names used for consume.
• topic #22: To improve feedback time, the topic information is now read and printed in parallel. This change breaks backwards compatibility as the output is not guaranteed to be sorted anymore (sort should be your friend here).
• kt #21: kt -help now prints build version and time information when built via make.

Jurek

Adds batching for the produce command, cf #18.

Inagaki

Adds functionality to set default values for -brokers and -topic via environment variables (cf. #14). Should reduce typing for shell sessions where you know the cluster or topic in advance. Values supplied via the command line take precedence over values from environment variables.

Herron

Fixes several edge cases around offset parsing. The following are supported now:

-1-           This will start consuming at the latest offset for all partitions
-1:100-       This will start consuming at offset 100 for all partitions
-1:30-,5:40-  This will start consuming at offset 30 for all partitions except 5 for which it will start at offset 40.

Internal:

• Adds setup for testing consume functionality more thoroughly.

Gebrselassie

UI:

• consume: Allows to specify offsets per partition (cf #3).
• produce: Allows to specify key, value and partition (cf #17).
• topic: Sorts topics alphabetically (cf #13). Thanks @echojc

Internal:

• Use godeps for vendoring (cf #16). Reduces repository's footprint. Thanks @marianogappa
• Updates to README and command usage information displayed for -help.

Breaking changes:

• consume: Drops text output and -json flag, defaults to JSON output (cf #15).

Fattore

• Breaks the UI of the topic command: lists by default now, option to filter by topic via regexp
• More optional information per topic/partition: leader and replicas

El Ouafi

• Adds a produce command (read lines from stdin) (#4)
• Adds a topic command that can list detailed partition info (thanks @echojc) (#5 and #11)

D'Haene

• Breaks existing UI by requiring a consume prefix for the consumer sub-command. This is mostly a start for adding more sub commands.
• Adds option to timeout consumer, cf #10.

Calcaterra

Defaults broker port to 9092 (cf #8, thanks @echojc)