The article is from Abhijit Menon-Sen, and its translation is from Thiago Belem. Both differs from the approach I used couple years ago, which involved having a full repo at server (let’s call it repo1), outside the public docs, and another bare repo at

/www

repo2

repo1

post-commit

The advantage of this method is that you could do little changes on the

/www

/www

But I admit, it’s a bit confusing. So both articles above have a much cleaner solution that I just tested and it’s working just fine. I won’t be able to vi the server, but that’s a compromise I’m comfortable with.

Step 1: Create the local GIT repository

$ mkdir -p ~/Sites/domain.com/www/; cd $_
$ echo Hello Dog > index.html
$ git init
$ git add *
$ git commit -am "inicial"

 Step 2: Prepare the server

ssh into the server, and make sure you have your RSA key on ~/.ssh/authorized_keys. If doesn’t… well, add it. :)

$ mkdir ~/domain.git; cd $_
$ git init --bare

Step 3: Create a post-receive hook on server

$ cat > hooks/post-receive << EOF
>
#!/bin/sh
GIT_WORKTREE=/home/www/domain.com git checkout -f" > a.txt
EOF
$ chmod +x hooks/post-receive;

Step 4: Configure local machine

$ git remote add www ssh://domain.com/home/domain/domain.git
$ git push www +master:refs/heads/master

5: Test the thing!

After the first commit, the normal flow would be:

• [change local file]

• git commit -am "Always comment your commits"

• git push www

Bonus Track: Different Environments

If you have a real website, with billions of billions of users, then you’ll need to test your changes before you push them live (unless you’re like that XX guy).

In that case, just add remotes pointing to different repositories, and repeat the same steps above to configure it. Then you’ll just push to different repos by doing:

git push testing
git push www

Phishing é uma forma de fraude eletrônica, caracterizada por tentativas de adquirir dados pessoais de diversos tipos; senhas, dados financeiros como número de cartões de crédito e outros dados pessoais. (http://pt.wikipedia.org/wiki/Phishing)

No dai 1 de maio, recebi um email da “Receita Federal”. De tão simples até parecia legítimo. Eis o email:

Levou aproximadamente meio segundo para pensar que era treta. Mas resolvi ler o email, vai que não é…

A linguagem pareceu adequada. O visual simples. O link não pareceu tão absurdo. Claro que tem umas coisas que facilitam, por exemplo eles colocaram um número qualquer da declaração, não mostram nenhum dado pessoal além do email. Aí fui ver o site que estava hospedando a página de coleta de dados:

Mandei email para a empresa informando que estavam sendo usados para tentativa de fraude e procurei na receita um email para copiá-los, mas não tem. De qualquer forma, no site da Receita eles estão avisando que existem várias tentativas, então já devem estar cientes.

É preciso ter cuidado… sempre.

All required node modules are embed into the plugin, so no need to npm anything.

How to install:

1. Be sure to have the awesome Package Manager from wBond installed. I’ve already submitted the package, but might take some time to appear on the list.
2. Open Sublime and go to Preferences > Browse Packages. Open the folder User, and then a terminal window on that folder. On Windows I’m using Git Bash Console window.
3. $git clone https://github.com/bitbonsai/JSHint-Inline.git
4. All done. Now when you have an HTML file with script> tags, just choose Tools > Build System > JSHint-Inline
5. And press CTRL + B. Result?

I consider myself a rather eclectic guy when the subject is OSs. At home I have OSX, Windows7 and Ubuntu 12.10, and what I really like is the fact that I use the same code editor for all of them (buy Sublime, the best editor ever) and that I have all servers installed on all of them. Ok, for Windows I downloaded binaries, but they work. Just git clone a website and run it, simple like that.

But then, I got that error above. So I’ll document the steps to make it work here, and save myself (and hopefully others) some time with this notes.

preReqs

Npm is pretty awesome and brainless, but it’s good to take a look at the error log files if a requisite is missing. I had to download Python (and installed python’s win32 extensions) Already have VS2012 installed, so no need for C++ runtime, and the most important: downloaded GNUWin32 Utils. 400 and something packages, look into the folder for .bat files and read them. The first to be executed is download.bat.

After all is set, I’ve tried npm -g install sqlite3; it installed, but can’t use, don’t know why. Then when I installed with npm install sqlite3 (without -g) it created a node_modules folder and placed the module inside it. That did the trick. Working like a charm!

UPDATE

What was missing to make it work installing with -g was NODE_PATH in Environment Variables. Right-click My Computer > Properties > Advanced system settings > Environment Variables and add:

NODE_PATH
C:\Users\YOUR_USERNAME\AppData\Roaming\npm\node_modules

To test it, a little script, save as nodesqlite3.js:

var sqlite3 = require('sqlite3').verbose();
var db = new sqlite3.Database(':memory:');

db.serialize(function() {
db.run("CREATE TABLE lorem (info TEXT)");

var stmt = db.prepare("INSERT INTO lorem VALUES (?)");
for (var i = 0; i < 10; i++) {
stmt.run("Ipsum " + i);
}
stmt.finalize();

db.each("SELECT rowid AS id, info FROM lorem", function(err, row) {
console.log(row.id + ": " + row.info);
});
});

db.close()

And the result?

$ node nodesqlite3.js
1: Ipsum 0
2: Ipsum 1
3: Ipsum 2
4: Ipsum 3
5: Ipsum 4
6: Ipsum 5
7: Ipsum 6
8: Ipsum 7
9: Ipsum 8
10: Ipsum 9

http://maps.googleapis.com/maps/api/js/GeocodeService.Search?4sporto%20alegre&7sbr&9sen-US&callback=_xdc_._t9nhk5&token=32861

And opening this on a new tab, checked this result:

_xdc_._t9nhk5 && _xdc_._t9nhk5( {
   "results" : [
      {
         "address_components" : [
            {
               "long_name" : "Porto Alegre",
               "short_name" : "Porto Alegre",
               "types" : [ "locality", "political" ]
            },
            {
               "long_name" : "Rio Grande do Sul",
               "short_name" : "RS",
               "types" : [ "administrative_area_level_1", "political" ]
            },
            {
               "long_name" : "Brazil",
               "short_name" : "BR",
               "types" : [ "country", "political" ]
            }
         ],
         "formatted_address" : "Porto Alegre - Rio Grande do Sul, Brazil",
         "geometry" : {
            "bounds" : {
               "northeast" : {
                  "lat" : -29.93063560,
                  "lng" : -51.0124710
               },
               "southwest" : {
                  "lat" : -30.26880690,
                  "lng" : -51.30614780
               }
            },
            "location" : {
               "lat" : -30.02770410,
               "lng" : -51.22873460
            },
            "location_type" : "APPROXIMATE",
            "viewport" : {
               "northeast" : {
                  "lat" : -29.93063560,
                  "lng" : -51.0124710
               },
               "southwest" : {
                  "lat" : -30.26880690,
                  "lng" : -51.30614780
               }
            }
         },
         "types" : [ "locality", "political" ]
      }
   ],
   "status" <code>: "OK"
}
 )

Checking the formatted_address part, and I have the full address. I can then record that into the DB, or that AND the lat/lng. VERY interesting… :)

json_decode(json_encode(simplexml_load_string(file_get_contents($xml_url))), 1);

Basically, I’m scraping Weather Channel’s local page, grep for the image tag and sed the src of that image, saving it to /tmp/weather.png. Then the geektool looks for that image and refreshes it every half an hour. The cron also checks for a new image in 30 minutes interval…

0,30 * * * * /Users/mw/bin/weather.sh

The weather.sh is also pretty simple:

curl --silent -o /tmp/poa.html "http://www.weather.com/weather/right-now/Porto+Alegre+Brazil+BRXX0186"; grep "<img class=\"wx-weather-icon\".*>" "/tmp/poa.html" | sed -n 's/.*src="\([^"]*\)".*/\1/p' | xargs curl --silent -o /tmp/weather.png

And the final result:

When I want to publish something, I want to write in markdown, save the file, commit and BAM! It’s published. I had high hopes on Posterous but then you have spaces, too many features… blablabla. And then you have email html parsing, never looks like I want or adds too much clutter…

I was looking into Jekyll, but then… I’m a PHP guy, would I be able to write something that just works? So I’m thinking about the features and maybe I’m gonna write something, why? Because yes.

• Markdown for posts/pages
• git fork publishing
• ftp publishing
• no admin hassle
• URL routes automagically
• Easy templates

I’ll update my progress, once I have time to start something so unnecessary…

So many interesting things, all claiming to be the next revolution in WebApp development… Right now seems that it’s the time to bet in one of them and see who wins the race.

I’m betting on Angular, the only thing that still bothers me is the tendency to create too many “reusable” components that are not plain HTML tags, like their <tabs> example on http://angularjs.org/

Also, they have a huge parter (Google) and on their first meetup they revealed that google is going to launch apps written in angular.