Support input arguments templating

[guoger]

today, tape only support static and fixed arguments upon start. Although this is way too limited. We need to pick and support a templating language. For example: https://ghz.sh/docs/examples#metadata-using-template-variables

[guoger]

cc @SamYuan1990

[wangxinalex]

How about the dynamic variables in postman? https://learning.postman.com/docs/writing-scripts/script-references/variables-list/. Some random data generator can be leveraged like https://github.com/brianvoe/gofakeit

[SamYuan1990]

How about the dynamic variables in postman? https://learning.postman.com/docs/writing-scripts/script-references/variables-list/. Some random data generator can be leveraged like https://github.com/brianvoe/gofakeit

well, my personal opinion doesn't limit for any kind of implementation.
for post man, well @wangxinalex could you explain more about how integrate post man with tape(a golang base program?)
for the random data generator is a good point. however, if we extend our view. the use case is that

tape user input something in config.yaml xxx word and during tape running replace xxx word with data.
for random case, if user wants random data, this library is very good for us.
@guoger , could you please help comparing the different between https://ghz.sh/docs/examples#metadata-using-template-variables and https://ghz.sh/docs/examples#metadata-using-template-variables ?

[wangxinalex]

How about the dynamic variables in postman? https://learning.postman.com/docs/writing-scripts/script-references/variables-list/. Some random data generator can be leveraged like https://github.com/brianvoe/gofakeit

well, my personal opinion doesn't limit for any kind of implementation.
for post man, well @wangxinalex could you explain more about how integrate post man with tape(a golang base program?)
for the random data generator is a good point. however, if we extend our view. the use case is that

tape user input something in config.yaml xxx word and during tape running replace xxx word with data.
for random case, if user wants random data, this library is very good for us.
@guoger , could you please help comparing the different between https://ghz.sh/docs/examples#metadata-using-template-variables and https://ghz.sh/docs/examples#metadata-using-template-variables ?

In my view, the dynamic variables in postman could provide some insights for tape due since they are both popular request generator. It is unnecessary to port these features from postman 100% because it is based on Javascript, instead we can try to leverage fake data generators like https://github.com/brianvoe/gofakeit to produce some fake data with semantic meaning.

On the other hand, ghz can replace templates with some contextual data https://github.com/bojand/ghz/blob/06f029c14c97612a09095f55075474442ffc8243/runner/calldata.go#L24, of course which can be seen as some random info source, however this may be insufficient if we want semantically meaningful payload.

[guoger]

How about the dynamic variables in postman? learning.postman.com/docs/writing-scripts/script-references/variables-list. Some random data generator can be leveraged like brianvoe/gofakeit

this looks promising.

Suppose following scenario:

1. create 1M accounts with random initial balance, then
2. randomly transfer among those accounts

1 should be easy with dynamic variable. Although I'm scratching my head around 2, where accounts are inherited from 1.

Anyway, satisfying 1 should be a good starting point already, and we could always build on top of it.

@wangxinalex interested?

[wangxinalex]

How about the dynamic variables in postman? learning.postman.com/docs/writing-scripts/script-references/variables-list. Some random data generator can be leveraged like brianvoe/gofakeit

this looks promising.

Suppose following scenario:

1. create 1M accounts with random initial balance, then
2. randomly transfer among those accounts

1 should be easy with dynamic variable. Although I'm scratching my head around 2, where accounts are inherited from 1.

Anyway, satisfying 1 should be a good starting point already, and we could always build on top of it.

@wangxinalex interested?

Sure, let's start with simple semantic random variables.