Add Pedestal (Clojure) about web-frameworks HOT 7 CLOSED

@kalicki why closed ?

from web-frameworks.

@waghanza I think something in Clojure would be cool, but I don't have time to implement something at the moment =(

from web-frameworks.

These are still good ideas though for someone to implement (which is always a good task to learn a language/framework too!). :-)

from web-frameworks.

Hi @redinger @ddeaguiar,

Would you mind to add (or consent that we add) pedestal in here ?

This project compare performance based on 3 routes :

• GET /, SHOULD return a successful status code and an empt²y string
• GET /user/:id, SHOULD return a successful status code and the id
• POST /user, SHOULD return a successful status code and an empty string

Regards,

from web-frameworks.

Hi @waghanza -- Very commendable work here pulling all these languages together in a single benchmark! I particularly don't like chasing benchmark games because I think they're disingenuous to most users and I don't have a lot of spare time to keep on another benchmark (I keep an eye on the TechEmpower ones). It's not clear what demonstrable value the benchmark has -- eg: What does this benchmark really tell you?
There's also another version of Pedestal being used/developed internally in some production systems that performs over an order of magnitude better than the current version (in goodput and latency) -- our attention is mostly focused on getting that ready to be rolled back into the open source project.

I'd prefer Pedestal to be left out of the benchmark, but you're free to do whatever you'd like.
Some tips if people attempt to do this on their own:

• Use a custom chain-provider (tailored to whatever platform you want to run on).
• Use static byte-buffer/byte-array responses to minimize GC churn
• Don't adhere to the Ring-spec (it's memory inefficient), just code directly to the platform you're on
• When you profile, pay particularly close attention to eden-generation GC churn
• Use a custom interceptor stack, don't use the default interceptors (you won't need them in the benchmark)
• Test various router setups. You might find with such a small set of routes, a simple cond will out perform everything (especially if no one else is using an actual router)

from web-frameworks.

HI @waghanza, thanks for reaching out! I agree with what @ohpauleez said.

from web-frameworks.

Hi @ddeaguiar @ohpauleez,

I understand the argues employeed. This benchmarks is not about making any competition between languages or frameworks.

I think, the more valuable here is to display some indicators (performances for now, but consumption will come) and to show how the frameworks could be written.

The implementation are simple (we take care of this), and useful for both decision makers and framework's author

Why for decision makers ?

Show indicators (a fair point is that this project is still young, and we will improve)

Why for frameworks author ?

Showing indicators could lead to questions about a more efficient implementation

We are aware that the purpose of this project could be mis-understood, but we are here to help ❤️

PS : We (at least I) won't accept any contribution that add pedestal if you are not giving you consent. I think it could be totally understandable if you do not want pedestal here ❤️

from web-frameworks.