[erlang-questions] Suggestions for easy to use super stable lightweight http server?

[Max Bourinov]

Hi Erlangers,

I am looking for easy to use super stable lightweight http server that supports https. I want to use it for facebook credit stuff integration. It won't be intended to serve normal http traffic like pages etc. I have to be able to configure app path, so it will be called like this: http(s)://host:port/some/path/here/

I thought about webmachine since we use it, but it is too much for my task. Any other suggestions with links?

Best regards,

Max

[Joel Reymont]

Did you look at built-in httpd?

---
Hardcore embedded Linux development
http://www.linkedin.com/in/joelreymont
Device drivers, firmware and debugging

> _______________________________________________
> erlang-questions mailing list
> erlang-q...@erlang.org (mailto:erlang-q...@erlang.org)
> http://erlang.org/mailman/listinfo/erlang-questions

_______________________________________________
erlang-questions mailing list
erlang-q...@erlang.org
http://erlang.org/mailman/listinfo/erlang-questions

[Max Bourinov]

Thank you for your feedback.

I think I will stick with: https://github.com/ostinelli/misultin

Best regards,

Max

[Zabrane Mickael]

On Dec 21, 2011, at 2:17 PM, Max Bourinov wrote:

Thank you for your feedback.

I think I will stick with: https://github.com/ostinelli/misultin

I STRONGLY suggest you cowboy:

https://github.com/extend/cowboy

A rock solid "Small, fast, modular HTTP server and socket acceptor pool"

[Max Bourinov]

Yeh... Cowboy sounds good, but it has two days old commits on http request functions. This seems not very cool to me... I really afraid to take it to production.

Best regards,

Max

[Max Bourinov]

And again, after deep consideration webmachine won.

Best regards,

Max

[Zabrane Mickael]

On Dec 21, 2011, at 2:34 PM, Max Bourinov wrote:

Yeh... Cowboy sounds good, but it has two days old commits on http request functions.

Cowboy is actively maintained, that's why.

I don't think the changes will break anything.

This seems not very cool to me... I really afraid to take it to production.

Maybe Essen (cowboy's author) can comment on that.

Regards,

Zabrane

[Tom Burdick]

I've been using cowboy for close to 3 months now for a project.
Nothing I've written has had to change. That doesn't mean there is any
guarantee things won't change (it isn't 1.x yet after all right?) but
I'd say cowboy's api is relatively stable at this point. Loic would
know best though!

If you like webmachine cowboy has the rest logic from webmachine in it
now written in plain english. See cowboy_http_rest.erl. You can pretty
much take the webmachine documentation and apply it to cowboy with a
few differences. Its also about 2-3x faster in my experience.

-Tom

> _______________________________________________
> erlang-questions mailing list
> erlang-q...@erlang.org

> http://erlang.org/mailman/listinfo/erlang-questions

>
_______________________________________________
erlang-questions mailing list
erlang-q...@erlang.org

http://erlang.org/mailman/listinfo/erlang-questions

[Samuel Elliott]

An article I wrote a while ago (but I believe it's still current)
enumerating erlang's http library offerings. I hope that it is useful:
http://lenary.co.uk/erlang/2011/08/erlang-web-libraries/

Sam

--
Samuel Elliott
s...@lenary.co.uk
http://lenary.co.uk/
+44 (0)7891 993 664

[Heinz N. Gies]

I can double that,
cowboy is pretty stable, the changes in the file were AFAIK only improvements to the interface that extended the functionality not change it. Also that said I was using misultin before (admitedly a pretty old version like 6 month old) and updating broke my entire code to a point where migrating was less effort.
--
Heinz N. Gies
he...@licenser.net
http://licenser.net

[Roberto Ostinelli]

hi,

misultin author here. misultin has *never* broken backwards compatibility.

can you please support such a harsh statement with an example on how upgrading "broke your entire code"? that might be helpful.

thank you.

r.

[Max Bourinov]

Hi Roberto,

That statement was not about misultin but about our approach in general. Please, don't take it personaly. Actually I like API of misultin.

We will use webmachie only because we have experience with it. So, if something will go wrong in one place something similar should be expected everywhere and if everything is alright there are much less chances that something will go wrong because of less code involved in the service delivery process. This is what I believe in.

So, I promise if there are any problems with web interface in my project, misultin will be the man and I will write about in this mail list.

Sent from my iPhone

_______________________________________________

[Roberto Ostinelli]

On Wed, Dec 21, 2011 at 9:46 AM, Max Bourinov <bour...@gmail.com> wrote:

Hi Roberto,

That statement was not about misultin but about our approach in general. Please, don't take it personaly. Actually I like API of misultin.

We will use webmachie only because we have experience with it. So, if something will go wrong in one place something similar should be expected everywhere and if everything is alright there are much less chances that something will go wrong because of less code involved in the service delivery process. This is what I believe in.

So, I promise if there are any problems with web interface in my project, misultin will be the man and I will write about in this mail list.

Hi Max,

I think you misunderstood me, I was not referring to your email. A library choice is obviously in developer's hands, who choose one for its features, performance, code usability, and such. I think only good things of webmachine, mochiweb, yaws, and cowboy.

Instead, I was referring to Heinz [snippet of his email]:

On Wed, Dec 21, 2011 at 7:45 AM, Heinz N. Gies <he...@licenser.net> wrote:

 I was using misultin before (admitedly a pretty old version like 6 month old) and updating broke my entire code to a point where migrating was less effort.

I've dedicated a lot of attention in not breaking backwards compatibility on misultin. Since AFAIK Heinz did not file any issues regarding this (I do realize that actually contributing to an open source project is much more time consuming than just plain stating opinions on a mailing list), I'm just curious to have some feedback on how that could happen, should Heinz be willing to share that information with me.

I just don't like receiving public criticism of my work in this way, as I always hope to get constructive criticism. Otherwise, I might as well just keep my source closed, for that matters.

r.

[Max Bourinov]

Hi Roberto,

I agree with you. I know what does it involves to manage open source project. I do really appreciate with you do for community. 

Best regards,

Max