Não foi possível carregar o Diqus. Se você é o moderador, por favor veja o nosso guia de problemas.

Marcos Bérgamo • 11 anos atrás

You maybe can adding a comparative using Node.js too?

Henrique Bastos • 11 anos atrás

Cool post, Avelino! It triggered another line of though I would like to share.

The problem with this kind of analysis is that what is being measured is explicit (requests/second) but what is being valued is implicit (computations/second).

Considering the above as truth, yes, you're right.

What I like to add to this discussion is a critic about thinking that "the best tool for the job" equals "the tool which gives you a better computations/second ratio".

When we take a big picture of the everyday life of most professional programmers, what is actually valued is "value/money". For most of the projects, development time and salaries are far more relevant constraints than how many VPS instances it will require to run.

By choosing to optimize code without making sure (with numbers) it's aligned with the project's profitability, one might be actually hurting his own sustainability.

0.01% of performance improvement on request handling at Google can save you enough money to hire a bunch of new developers. 10% of the same performance improvement in a small company may not even enable you to improve the team's breakfast.

So I think the "best tool" is that good enough for you to deal effectively with *all* the constrains that you have (not only computational ones).

With that in mind, I see the Python (my personal preference) ecosystem with all it's tangible and intangible awesomeness as a far better tool, accessible to anyone wanting to make it happen.

Avelino • 11 anos atrás

I totally agree! I really like Go and am extremely passionate about Python, is a language that makes me be productive with quality, I commented at the end of the blogpost that we need to put on the table what we need for the application we're developing and that there is no Ballad of silver, who wants language we're going to win on one side and losing another.

Gustavo Gawry • 11 anos atrás

You should test with a faster wsgi server for python like bjoern.

Avelino • 11 anos atrás

Really improved a lot, the following results:
source: https://gist.github.com/ave...
result: https://gist.github.com/ave...

Guest • 11 anos atrás

See if we can beat Go using Wheezy.web (https://pythonhosted.org/wh...

Gustavo Gawry • 11 anos atrás

Try now with Wheezy.web instead of Falcon, and see if we can beat Go net/http. :-)

(https://pythonhosted.org/wh...