The async'd non-block'd server
I have decided to abandon this project and to team up with the guys of http://www.nidium.com/.
These two projects have a lot in common:
- Goal wise:
- networking async non blocking
- high performance eventloop
- embedded javascript
- lots of modules out of the box
- ..
- profit
- Technology wise:
- Speed from C/C++
- adapting cool libraries
- embedded Spidermonkey from Mozilla
- ...
- fun
Do not fear:
- as long as https://github.com/nidium/ is not there, we are working hard on cool stuff :-)
- as long as https://github.com/verpeteren/urbin is here, it is not lost... :-)
Let's face it. Web-apps are here to stay for the next decade, and we all want to be super-productive.
The browser side is javascript, using another language on the server causes too much thinking. So the serverside should be javascript as well.
I always felt that Node.js and vert.x were too magic and I do not want to land in mediocracy, so I do not want to follow the mainstream.
I want to have a special server, that is :
* fast webserver
* scriptable with javascript
* that can talk with postgresql
* works asyncronously
* is blazing fast
* is super stabile
* extendable with modules other then webserver, a sqlclient
Building upon the good ideas from others, glue it together.
stability: compile with full warnings / valgrind all the way
Fast: non blocking, async, zero copy
Meme: KEYMAKER: But like all systems, it has a weakness. The system is based on the rules of a building, one system built on another
Something cool
Peter Reijnders [email protected]
#Create some platform specific settings, gets the dependencies and compiles `./build.sh
#Get and compile the dependencies (may take some time)
`cd ./src
make deps
#Then compile
make all
#Then configure
cd ../bin
vi ./etc/urbin.conf
#then hack
vi ../var/scripts/javascript/main.urbin.js
#Then run
./urbin
- Following ID software's code style guide: http://www.geeks3d.com/downloads/200811/idSoftware-Coding-Conventions-Geeks3D.com.pdf
- Take track of memory alloctions and the appropiate cleanup with an anonymous struct see http://blog.staila.com/?p=114 for details
These tools are needed to compile the dependencies to build, but after that, they are not needed anymore.
sudo apt-get install autoconf2.13 wget git sed unzip build-essential g++ make cmake python2.7 head tail cat tac ar ranlib tr echo cut basename dirname echo strip jsdoc-toolkit
| Topic | Project | License | SLOC | Link |
|---|---|---|---|---|
| Eventloop | picoev | BSD | 1140 | http://developer.cybozu.co.jp/archives/kazuho/2009/08/picoev-a-tiny-e.html |
| Http Parser | H3 | MIT | 481 | https://github.com/c9s/h3 |
| Logging | c-logging | Public Domain | 230 | https://github.com/dhess/c-logging |
| Dns | TADNS | Beer-ware | 570 | http://adns.sourceforge.net/ |
| Url | urlparser | BSD | 5050 | http://uriparser.sourceforge.net/ |
| Regex | oniguruma | BSD | 19458 | http://www.geocities.jp/kosako3/oniguruma/ |
| Postgresql | pq | Postgresql | 53205 | http://www.postgresql.org/docs/9.4/static/libpq.html |
| Portabilty | nspr | MPL2.0 | 112077 | https://developer.mozilla.org/en-US/docs/Mozilla/Projects/NSPR |
| Javascritp | spidermonkey | MPL2.0 | 439718 | https://developer.mozilla.org/en-US/docs/Mozilla/Projects/SpiderMonkey |
Topic Project License SLOC Link Mysql mysac GPL3 18593 http://cv.arpalert.org/page.sh?mysac
To download and compile the dependecies the deps directory is a little bit over 3.5 Gb! Amazing..
TBD, probably MIT
MySql can be excluded from the linking
Q: Why does this project use Makefiles?
A: They help with 'once-and-only-once' definitions
Q: Why do you download and compile all the dependencies yourself? You could use the system libraries.
A: As this is cutting edge, the dependencies have not landed in your distro yet, or are ancient.
Q: What is with the name?
A: Cool domain names are hard to get.
Q: What is the relationship with APE-Project?
A: Ajax-Push-Engine and libapenetwork are focussed on async sockets; This project strives to be less advanced (KISS) then APE. However some ideas appear to be simular.
Q: What is the relationship with H2O?
A: H2O is focussed on writing a very fast and complete HTTP1 and HTTP2 server. This project strives to be less advanced (KISS) then H2O. The project lead of H2O is the same genius that wrote the picoev event loop.
E: error while loading shared libraries: libXXX.so: cannot open shared object file: No such file or directory
S: export LD_LIBRARY_PATH=pwd``
E: ./etc/urbin.conf:2: "no such option" or "missing title for section" or "invalid integer value for option"
S: check that config file, in line 2: there is a typo