After installing those dependencies, delete $VIM_SOURCE/src/config.cache or there’ll be variables set on that file that’ll fool ./configure into thinking that those dependencies are still missing.

Fuck yeah!!

...
Undefined symbols:
  "_iconv_close", referenced from:
      _buf_write in fileio.o
      _readfile in fileio.o
      _readfile in fileio.o
...

The problem has to do with lib search path. I don’t know yet why isn’t used the system path (under /usr/lib) rather than the one used by macports.

Well, the thing is that the libiconv2.dylib under /opt/local/lib doesn’t have the _iconv_close function (among others):

[madtrick { lib } ]pwd
/opt/local/lib
 
[madtrick { lib } ]nm libiconv.2.dylib | grep "iconv_close"
000163f0 T _libiconv_close

While libiconv.2.dylib on /usr/lib does:

[madtrick { lib } ]pwd
/usr/lib
 
[madtrick { lib } ]nm libiconv.2.dylib | grep "iconv_close"
0000b905 T _iconv_close
0000fcce T _libiconv_close

So to fix this problem you have several choices:

• Deactivate libiconv on macports (and all its dependencies)
• Prepend a lib path

I went for the second one:

LDFLAGS="-L/usr/lib" ./configure _your_options_here_

There you go : )

Introducing: __git_ps1

Usage:

Include a call to __git_ps1 in your promt string (for example in bash):

PS1='[\u@\h \W$(__git_ps1 " (%s)")]\$ '

This will show the current branch on your prompt.

Customization:

__git_ps1 allows to customize its output using environment variables. Here is a list of avaliable options:

• GIT_PS1_SHOWDIRTYSTATE.  When set to something not empty, __git_ps1 will display unstaged (*) and staged (+) next to the branch name.
• GIT_PS1_SHOWSTASHSTATE.  When set to something no empty, __git_ps1 will display $ next to the branch name when something is staged.
• GIT_PS1_SHOWUNTRACKEDFILES.  When set to something no empty, __git_ps1 will display % next to the branch name when there’re untracked files.
• GIT_PS1_SHOWUPSTREAM.  This cat take several (comma separated) values: auto, verbose, legacy, git and svn. __git_ps1 will show the difference between HEAD and its upstream.

Export them if your shell dotfile before calling __git_ps1.

Until now…

Today I found TerminalTabSwitching. A plugin to enable cmd + <tab_number> switching on Terminal.app.

I’m a bit happier : )

Don’t set Jetty’s <stopPort> to the same <port> where Jetty is running or you might run into some nasty-hard-to-debug problem.

Now a full explanation:

Imagine the following scenario,

Here we have to web services, A and B, both running on the same machine on Jetty webserver and one of them is listening for petitions on port 9090 (configured via de maven-jetty-plugin). Web service A has to request some data to B and process its reply. Nothing unusual.

So A issue a request to http://localhost:9090/<path> expecting an XML response as a result of doing it. But it gets nothing, nada, nichts.

Well, at this moment the first thing that came through my mind was that I did something wrong while coding or configurating any of the web services. So I checked everything, but it seemed right. My next move was to create a fake request with curl just to see what happened.

What happened misleaded me even more: web service B was responding. WTF!!

So I went one level down. I ran tcpdump. And what I got? I discovered that the main difference between the requests issued by web service A and curl was the ip version they used: web service A was using ipv4 and curl ipv6.

To force both requests (from the web service and curl) to use ipv4, I run curl against the web service B but now instead of localhost I used 127.0.0.1. And the response was empty, just like if the request came from web service A.

So I thought that all of my troubles had to do with the ip stack beeing used. And I wasted some hours
trying to figure out what was wrong with it.
As I was really desperate with all this, I download Jetty sources to debug it on eclipse. I wanted to know
what was happening with my requests. And here was were light came to me by chance. While debugging Jetty,
an incomming connection from  internet arraived to my <public_ip>:9090  (I guess it was someone doing some kind of portscanning) and got stuck on my breakpoints…

Mmmm, wait a moment…

Yep, internet addresses are ipv4 only (for most of us), so this ruled out any kind of problem with the ipv4 stack or shit. Now I began to see the light. This was some mess with the adresses.

Just to check I did an alias in my loopback interface to 127.0.0.2 and run again the request with curl to that ip. This time being successful. Ooook.

So, whats wrong with 127.0.0.1:9090?

Netstat output,

13:29:46 ~ $ netstat -f inet -an
Active Internet connections (including servers)
Proto Recv-Q Send-Q  Local Address          Foreign Address        (state)
tcp4       0      0  127.0.0.1.9090         *.*                    LISTEN
tcp46      0      0  *.9090                 *.*                    LISTEN
...

Lsof output,

13:49:39 ~ $ lsof -i :9090
COMMAND PID     USER   FD   TYPE     DEVICE SIZE/OFF NODE NAME
java    239 madtrick   94u  IPv6 0x08c10f80      0t0  TCP *:websm (LISTEN)
java    239 madtrick   95u  IPv6 0x08c11460      0t0  TCP [::127.0.0.1]:websm (LISTEN)

I have two sockets listening on port 9090 and both belong to the same process (as reported by lsof), my Jetty Server.  But requests to 127.0.0.1:9090 are lost meanwhile requests to <any_other_ip>:9090 are processed by the server.

Lets try something, lets change the <stopPort> to 9091 and see what happens…
Request from web service A, OK.
Request from curl, OK.

Run netstat again,

13:59:50 ~ $ netstat -f inet -an
Active Internet connections (including servers)
Proto Recv-Q Send-Q  Local Address          Foreign Address        (state)
tcp4       0      0  127.0.0.1.9091         *.*                    LISTEN
tcp46      0      0  *.9090                 *.*                    LISTEN
...

And lsof for ports 9090 and 9091,

14:00:48 ~ $ lsof -i :9090
COMMAND PID     USER   FD   TYPE     DEVICE SIZE/OFF NODE NAME
java    301 madtrick   93u  IPv6 0x08c10f80      0t0  TCP *:websm (LISTEN)

14:00:55 ~ $ lsof -i :9091
COMMAND PID     USER   FD   TYPE     DEVICE SIZE/OFF NODE NAME
java    301 madtrick   94u  IPv6 0x08c11460      0t0  TCP [::127.0.0.1]:xmltec-xmlmail (LISTEN)

So you see, Jetty opens another socket which listens on 127.0.0.1:<stopPort>  (lets call it jetty stop daemon) and being that an explicit address it has precedence over the catch all 0.0.0.0:9090 (where Jetty is listening). That’s why my requests to 127.0.0.1:9090 returned nothing.

This is what was happening before,

And this is whats happens now,

BOSH, the technology defined in this specification, essentially provides a “drop-in” alternative to a long-lived, bidirectional TCP connection …

For applications that require both “push” and “pull” semantics, BOSH is significantly more bandwidth-efficient and responsive than most other bidirectional HTTP-based transport protocols and the techniques now commonly known as “Ajax”…

That is, BOSH is quite similar to WebSockets

So, at this point is where mochiweb comes in. It will provide the framework for handling the HTTP requests where the BOSH requests themselves are wrapped. But before going any further lets explain how mochiweb works, with a picture:

What do we have here?

1. An HTTP request comes in, to the address and port mochiweb is configured.
2. If it’s a valid HTTP request, a mochiweb_request “object” is created and the handler function at the callback module (which is provided when mochiweb starts) is invoked with this mochiweb_request as a parameter. Give a look to this and this to get a grasp at parameterized modules if you wish to know more about “objects” on Erlang.

Easy, huh?

And what is all that about a BOSH manager? The BOSH manager handles the connection between the resource-client (for example a browser) and the resource-owner (for example a server). It has to keep HTTP connections opened to fake this bidirectional comunication. That is, the client ask for something about the resource and so it sends a request to the BOSH manager, the BOSH manager sends this same request to the server and waits for a reply. If a reply arrives, then it’s forwarded to the client. But aside from this, which is no more than plain-old AJAX, BOSH offers PUSH technologies. If any event happens at the server and the client has to be notified, the BOSH manager forwards the event to the client. That’s why it has to keep HTTP connections opened, it must have a valid path.

In Erlang, creating and destroying processes is extremely cheap, so what I wanted was  to spawn a new process for each single BOSH session (between one client and one or multiple servers). Lets give another picture:

On the picture above, there are two separate BOSH managers (each of them is a Erlang process) BM1 and BM2 which handler connections from one client to one server (BM1 ) and from another different client to two servers (BM2)

To complete the setup I used, I have to talk about StropheJS which is a library for writing xmpp clients (BOSH is an xmpp extension).

Well, once I have explained more or less what I’m doing with mochiweb and BOSH, lets got to the details.

First, lets talk about my callback module for mochiweb. It’s not a ver big one nor a complicated one, not because I don’t know how to move on but was the error I’ll talk later which didn’t allow me to progress. So ladys and gentleman here is my litle callback module:

-module(mochibosh).

-export([start/0, start/1, stop/0]).
-export([handler/1]).

-export([init/0]).

-record(mochibosh_state,{procs, xml_models}).

handler(Req) ->
 io:format("CALLBACK. Petition on port ~p  ~n", [inet:port(Req:get(socket))]),
 mochibosh ! Req,
 ok.

start() ->
 erlang:register(mochibosh, spawn_link(?MODULE, init, [])),
 start([{ip,"192.168.1.110"}, {port, 8888}]).

start(Options) ->
 Options2 = [{loop, {?MODULE, handler}} | Options],
 mochiweb_http:start(Options2).

stop() ->
 mochibosh ! exit,
 mochiweb_http:stop().

% Internal API

init() ->
 Models = init_xml_models(),
 loop(#mochibosh_state{procs = dict:new(), xml_models = Models}).

loop(State = #mochibosh_state{procs = Procs, xml_models = Models}) ->
 receive
 Request ->
 io:format("SPAWNED PROCESS. Request received on port ~p ~n", [inet:port(Request:get(socket))]),
 loop(State)
 end.

init_xml_models()->
 [{bosh_init, init_xml_model("bosh_init.xsd")}].

init_xml_model(Xsdfile) ->
 {ok, Model} = erlsom:compile_xsd_file(Xsdfile),
 Model.

It very easy. If want to use it, you cant fire a Erlang terminal and type mochibosh:start() and it will be up and running. The loop
option in the tuple that is passed to mochiweb_http:start is used to provide the callback module. In this case is the mochibosh module itself (?MODULE)

So what would happen if we try it….

In one terminal we fire the Erlang shell (erl) and start mochibosh:

20:38:25 mochibosh $ erl
Erlang R14A (erts-5.8) [source] [smp:2:2] [rq:2] [async-threads:0] [hipe] [kernel-poll:false]

Eshell V5.8  (abort with ^G)
1> mochibosh:start().
{ok,<0.34.0>}

In another terminal we issue a simple request with curl…

20:40:19 mochibosh $ curl -v http://192.168.1.110:8888
* About to connect() to 192.168.1.110 port 8888 (#0)
*   Trying 192.168.1.110... connected
* Connected to 192.168.1.110 (192.168.1.110) port 8888 (#0)
> GET / HTTP/1.1
> User-Agent: curl/7.21.0 (i386-apple-darwin10.2.0) libcurl/7.21.0 OpenSSL/1.0.0a zlib/1.2.5 libidn/1.19
> Host: 192.168.1.110:8888
> Accept: */*
>
^C

And meanwhile in the Erlang shell..

{ok,<0.34.0>}
CALLBACK. Petition on port {ok,8888}
SPAWNED PROCESS. Request received on port {ok,8888}

2>

GREAT!! Everything seems to work. So now, I decided to move on and begin to develop my BOSH manager.

As I sayed, I’ll be using StropheJS for the browser/client side. For testing purpouses I began using one of the examples shipped with the library: echobot.js .This example will issue a request for starting a BOSH session (among other things) and that is perfect for my testing pourposes. So lets, try.

(Imagin that the request is sent from echobot to the BOSH manager)

And I expecto to get the same reply from my callback module as when I tryed it with curl. Lets see:

CALLBACK. Petition on port {ok,8888}
SPAWNED PROCESS. Request received on port {error,einval}

WTF!!!!! What’s that of {error, einval}. Well, I can see that’s an error, but why? As this post is getting really big, I’ll leave the explanation for another one.

1. I installed this on a MacBook with a CoreDuo processor. By default macports will try to use all avaliable processors for the building stuff (2 on my case) with the -j option of Make. On my case this ended up with weird compilation errors (something like a race condition between the to make jobs). I fixed it setting the buildmakejobs to 1 (at /opt/etc/macports/macports.conf) so it used just one make job
2. The Makefile rule for the documentation at <package>/support/include.mk is wrong. The function that make should call at the edoc module is files/3 not file as it’s deprecated.

Enjoy!!

Por defecto, IRB siempre “escupirá” el valor resultante de evaluar la expresión introducida por el usuario. Aunque la mayoría de las veces eso está bien, hay otras en las que sacarias el corazón con una cucharilla de Té  a quien tuvieras delante por la crispación que produce esta “verbosidad”.

Me refiero a situaciones del tipo “quiero recorrer un array y listar el nombre de sus elementos con puts”, ejemplo:

ArrayEnorme.each do |element| puts "#{element.name}\n" end

Con el código anterior vamos a mostrar el nombre de cada elemento del Array. Hasta aquí todo normal. El problema de evaluar esto en IRB es que, si ArrayEnorme tiene bastantes elementos y cada elemento tiene bastantes atributos, las lineas que imprimos con el nombre de cada elemento quedaran “sepultadas” por un montón de datos inutiles. Es decir, si el buffer de nuestra consola no se ha llenado con estos datos inutiles (perdiendo así las lineas con el nombre de cada elemento) tendremos que hacer scroll ad-infinitum, que para el caso es casí como si se hubierán perdido.

Para evitar esto, os presento mi solución de andar por casa:

def toggle_output
context.return_format == "Output canceled\n" ? IRB.conf[:PROMPT][context.prompt_mode][:RETURN] : context.return_format = "Output canceled\n"
end

alias tgo toggle_output

Después de copiar las lineas anteriores en nuestro .irbrc podemos hacer pruebas:

madtrick::madtrick::$irb
irb(main):001:0> 2 + 1
=> 3
irb(main):002:0> toggle_output
Output canceled
irb(main):003:0> 2 + 1
Output canceled
irb(main):004:0> toggle_output
=> "=> %s\n"
irb(main):005:0> 2 + 1
=> 3

Y voila, ya podemos callar y dar voz a IRB cuando nos plazca.

El concepto que implementa sketches es sencillo: las modificicaciones que hagamos en nuestro editor de texto, se reflejan al instante en la sesión de irb. Es decir, si en el editor de texto creamos una clase y guardamos los cambios, la clase estará disponible automaticamente en la sesión de irb.

Su utilización es sencilla: arrancamos el IRB y lanzamos nuestro editor favorito con el comando sketch. A partir de ahí, el funcionamiento es como expliqué antes.

Mola : )