This fizes a Ruby warning generated when running the tests for
Rackstash::Adaptes::Logger. The warning was:
> warning: instance variable @closed not initialized
The Sink was a vehicle to transport some shared state between the logger
instance and the buffers, most notably the default fields and default
tags.
It turns out however, that error handling during merging of the
default_fields and default_tags is non trivial since there, the buffer
is in sime kind of limbo: users won't write to it anymore (and thus
don't expect exceptions there) while the error handling of the
individual flows is not yet reached. Since users can specify procs in
default_fields, they can still raise for whatever user-defined reason.
Luckily, the insertion of default fields and default tags can easily be
done by a filter later anyway, under the protection of the flow's error
handling in Flow#write. This allows us just remove the whole concept of
the sink and just pass the Flows object to the Buffer.
Not having to merge default_fields during event creation significantly
simplifies Buffer#to_event which was rather ugly to begin with but now
turned out quite beatifully.
If messages are removed by the final cut step (but not when applying the
selectors), we insert an ellipsis, by default a Message with the
contents of "[...]\n", at the place where we removed messages. This
helps the investigating human who reads the logs to understand that
there were logs messages removed from the event.
This can be used to achieve similar results as
ActiveSupport::TaggedLogger. With our implementation, the "tags" are
prepended to each line (instead of each message) and we add the same
tags to all message lines.
If required, you can still use ActiveSupport::TaggedLogger with
Rackstash to add tags at the time of adding the log message.
If we have a proc, we can pass it directly to the `select!` method
instead of having to manually call the selector. This gives a
significant speedup for the common case of simple selectors:
require 'benchmark/ips'
Benchmark.ips do |x|
selector = ->(value) { value < 500 }
values = (1..1000).to_a
x.report("call") { values.select { |v| selector.call(v) } }
x.report("block") { values.select(&selector) }
x.compare!
end
Warming up --------------------------------------
call 663.000 i/100ms
block 1.598k i/100ms
Calculating -------------------------------------
call 6.897k (± 4.8%) i/s - 34.476k in 5.011617s
block 16.561k (± 4.2%) i/s - 83.096k in 5.026999s
Comparison:
block: 16560.7 i/s
call: 6896.5 i/s - 2.40x slower
Any public method is part of the public interface of the class and thus
needs to be equally well supported. Only because methods are not usually
called by end-users, we still need to properly document, test, and check
them.
This resolves the following warning:
> spec/rackstash/encoders/logstash_spec.rb:18: warning: ambiguous first
> argument; put parentheses or a space even after `/' operator
This allows to configure the last (and often only) flow on creation of
the Logger without having to manually create the object. We can thus use
the following shortcut to e.g. set a custom encoder:
Rackstash::Logger.new(STDOUT) do
encoder Rackstash::Encoder::Message.new
end
That way, we ensure that the error_flow receives an expected event and
doesn;t have to deal with different data formats which might result in
additional (then quietly hidden) errors.
