Use span directly instead of reader / writer for methods & properties

[bollhals]

Proposed Changes

1. Introducing a benchmark project to store actual benchmark code.
2. Replace the ContentHeaderProperty(Reader/Writer) & MethodArgument(Reader/Writer) by directly using the span.

Types of Changes

• Bug fix (non-breaking change which fixes issue #NNNN)
• New feature (non-breaking change which adds functionality)
• Breaking change (fix or feature that would cause an observable behavior change in existing systems)
• Documentation improvements (corrections, new content, etc)
• Cosmetic change (whitespace, formatting, etc)

Checklist

• I have read the CONTRIBUTING.md document
• I have signed the CA (see https://cla.pivotal.io/sign/rabbitmq)
• All tests pass locally with my changes
• I have added tests that prove my fix is effective or that my feature works
• I have added necessary documentation (if appropriate)
• Any dependent changes have been merged and published in related repositories

Further Comments

Results:
WireFormatting_Read_BasicAck

Method	Mean	Error	StdDev	Ratio	Gen 0	Gen 1	Gen 2	Allocated
ReadArgumentsFrom_MethodArgumentReader	6.949 ns	14.0979 ns	0.7728 ns	1.55	0.0068	-	-	32 B
ReadFromSpan	4.486 ns	0.2777 ns	0.0152 ns	1.00	0.0068	-	-	32 B

WireFormatting_Read_BasicDeliver

Method	Mean	Error	StdDev	Ratio	Gen 0	Gen 1	Gen 2	Allocated
ReadArgumentsFrom_MethodArgumentReader	23.91 ns	0.634 ns	0.035 ns	1.80	0.0119	-	-	56 B
ReadFromSpan	13.28 ns	3.272 ns	0.179 ns	1.00	0.0119	-	-	56 B

WireFormatting_Read_ChannelClose

Method	Mean	Error	StdDev	Ratio	Gen 0	Gen 1	Gen 2	Allocated
ReadArgumentsFrom_MethodArgumentReader	14.261 ns	0.6648 ns	0.0364 ns	1.70	0.0068	-	-	32 B
ReadFromSpan	8.343 ns	0.3412 ns	0.0187 ns	1.00	0.0068	-	-	32 B

WireFormatting_Write_BasicAck

Method	Mean	Error	StdDev	Ratio	Gen 0	Gen 1	Gen 2	Allocated
WriteArgumentsTo_MethodArgumentWriter	5.110 ns	0.2012 ns	0.0110 ns	2.47	-	-	-	-
WriteArgumentsTo	2.066 ns	0.0219 ns	0.0012 ns	1.00	-	-	-	-

WireFormatting_Write_BasicDeliver

Method	Mean	Error	StdDev	Ratio	Gen 0	Gen 1	Gen 2	Allocated
WriteArgumentsTo_MethodArgumentWriter	45.72 ns	0.908 ns	0.050 ns	1.15	-	-	-	-
WriteArgumentsTo	39.71 ns	1.600 ns	0.088 ns	1.00	-	-	-	-

WireFormatting_Write_ChannelClose

Method	Mean	Error	StdDev	Ratio	Gen 0	Gen 1	Gen 2	Allocated
WriteArgumentsTo_MethodArgumentWriter	19.72 ns	1.575 ns	0.086 ns	1.29	-	-	-	-
WriteArgumentsTo	15.24 ns	0.138 ns	0.008 ns	1.00	-	-	-	-

WireFormatting_Read_BasicProperties

Method	Mean	Error	StdDev	Ratio	Gen 0	Gen 1	Gen 2	Allocated
ReadFrom	96.78 ns	3.470 ns	0.190 ns	1.35	0.0408	-	-	192 B
ReadFromSpan	71.50 ns	0.860 ns	0.047 ns	1.00	0.0408	-	-	192 B

WireFormatting_Write_BasicProperties

Method	Mean	Error	StdDev	Ratio	Gen 0	Gen 1	Gen 2	Allocated
WriteProperties	63.90 ns	3.239 ns	0.178 ns	1.74	-	-	-	-
WritePropertiesToSpan	36.68 ns	0.331 ns	0.018 ns	1.00	-	-	-	-

From the test app:
Before:

After:

[michaelklishin]

@bollhals thank you, this looks reasonable. Can you please rebase this on top of master?

[michaelklishin]

@bollhals if this could go into 7.x, would it be possible to submit a version of this based on that branch? If anything, these changes would make it easier to backport other changes to 7.x: the delta with master keeps growing.

[bollhals]

@bollhals if this could go into 7.x, would it be possible to submit a version of this based on that branch? If anything, these changes would make it easier to backport other changes to 7.x: the delta with master keeps growing.

Sure that's possible. I'm wondering is there any change on master that is really 8.0 only? Or couldn't we merge master to 7.0?