First, let's make sure we have node
and npm
installed:
$ node -v
v0.10.25
$ npm -v
1.3.24
Now let's try to run gulp
. If everything goes to plan, we should see our first error:
$ gulp
command not found: gulp
Perfect! Let's ask NPM for a fresh copy of gulp
:
$ npm install gulp -g
... installing gulp ...
This installs gulp
command-line utility and makes available globally. Now, let's try running gulp
again:
$ gulp
[gulp] No gulpfile found
Nice, our first error!
gulp
is telling us that it doesn't have a gulpfile
with instructions for it to follow. We solve this by simply adding gulpfile.js
to our directory root:
$ touch gulpfile.js
We haven't added any instructions but sipmly having having the file should give us a new error:
$ gulp
[gulp] Using gulpfile /Users/chan/code/talks/gulpfile.js
[gulp] Task 'default' was not defined in your gulpfile but you tried to run it.
[gulp] Please check the documentation for proper gulpfile formatting.
Boom! Now we're talking. gulp
is running and it sees our gulpfile
. We're on a roll!
It tells us that we don't have a default
task. Every gulpfile needs a default
task to run, when gulp
is executed without arguments.
We find the syntax for creating a gulp
task in their docs. Looks like it's gulp.task('taskName');
. Let's try it:
$ echo "gulp.task('default');" >> gulpfile.js
$gulp
... ReferenceError: gulp is not defined ...
Yuck! Now we have a really ugly error...
But don't lose heart. This error looks scarier than it is.
node
is telling us that there is no reference to 'gulp' in our file. We simply need to add a reference to gulp
in our gulpfile
:
$ open gulpfile.js
[ add the following line to the very top of your gulpfile
]
// gulpfile.js
var gulp = require('gulp');
Now that we've told node what gulp
means, in our script, lets try running gulp
again:
$ gulp
[gulp] Using gulpfile /Users/chan/code/talks/gulpfile.js
[gulp] Starting 'default'...
[gulp] Finished 'default' after 71 μs
Hurray! We now have a working gulpfile
that does nothing! Pat yourself on the back firmly.
Now, gulp
just sits on top of node
. So, we can do anything we could do node:
// gulpfile.js
// update our default task with a function
gulp.task('default', function () { console.log('Hello Gulp!') });
Run this and we'll see Hello Gulp!
amidst the gulp
output:
$ gulp
[gulp] Using gulpfile /Users/chan/code/talks/gulpfile.js
[gulp] Starting 'default'...
Hello Full Stack
[gulp] Finished 'default' after 137 μs
This is particularly unuseful. So, let's turn this into something that can compile CoffeeScript into JavaScript.
Let's start with a new task for this: gulp scripts
.
We're going to mimic what we did with our default
task. gulp
gives great errors. It's easy to start with the process you want and respond to the errers until it works.
Let's try gulp scripts
$ gulp scripts
[gulp] Using gulpfile /Users/chan/code/talks/gulpfile.js
[gulp] Task 'scripts' was not defined in your gulpfile but you tried to run it.
[gulp] Please check the documentation for proper gulpfile formatting.
As I'm sure you expected, this fails. We haven't written a scripts
task. Let's write it now.
[ add this to your gulpfile
]
// gulpfile.js
gulp.task('scripts', function () {
gulp.src('src/*.coffee').pipe(gulp.dest('./'));
});
There's a lot of code we haven't covered here. So take a second to look it over. It may seem jarring at first but give it a second to sink it...
Now we can talk about what it's doing.
First, we've registered a new task
named scripts
. Much like our default
task, it responds by running a function.
That function start by calling gulp.src()
to collect the files in the path we specify. Here, we're using *
to say "grab any and all files in the src
directory.
We then use pipe()
to funnel those files through another process, gulp.dest()
.
gulp.dest()
places the files specified with a path. In example, we're using ./
to tell gulp that we want our finished files dumped into the root directory.
Notice how easily that reads? If you've been using JavaScript for a while, this should look painfully obvious.
Any ideas what will happen when we run this?
$ gulp scripts
[ this section requires some previous setup. Add file src/fun.coffee
for the examples to work ]
Gulp found our src/fun.coffee
file and processed it by creating a new fun.coffee
file in the directory root.
Great! But this isn't exactly what we want. All we've done is duplicate our file in the directory root.
We need to process the CoffeScript file before placing it in the destination path.
We can accomplish this by chaining another pipe()
right between gulp.src()
and .pipe(gulp.dest(...))
. This pipe
will handle the actual processing of the file.
[ update the task to look like this ]
// gulpfile.js
gulp.task('scripts', function () {
gulp.src('src/*.coffee')
.pipe(coffee())
.pipe(gulp.dest('./'));
});
Now, when we run this, you shold have a pretty good idea what's going to happen:
$ gulp scripts
[gulp] Using gulpfile /Users/chan/code/talks/gulpfile.js
[gulp] Starting 'scripts'...
[gulp] 'scripts' errored after 5.17 ms coffee is not defined
That's right. Another error!
You'll notice that gulp is getting smarter and our errors are less scary. gulp
tells us that we have another reference error. We need a coffee
processor. Let's add the reference to our gulpfile
:
// gulpfile
var coffee = require('gulp-coffee');
Now it should work, right?!
$ gulp scripts
Error: Cannot find module 'gulp-coffee'
AHHH! Our error is getting uglier! We have our reference but node
can't find gulp-coffee
.
To add it. We turn back to our trusty package manager, NPM.
$ npm install gulp-coffee
... installing gulp-coffee ...
And now, running gulp scripts
, we should get a fun.js
file placed in our root directory:
$ gulp scripts
$ ls
fun.js gulpfile.js src
"And there was much reqoicing."
It's worth pointing out that this isn't any better than using the coffee
cli. If we have to run the gulp
command each time we change our file, howe does gulp help us?
We need a way to watch the file system. And gulp.watch()
helps us with exactly that.
Like gulp.task()
, gulp.watch()
can take a path as it's first argument and a function as its second. But when we watch a directory, we may want to run a number of tasks!
gulp
accounts for this by allowing you to send an array of task names as the second argument. Let's create a new task watch
to play with this:
// gulpfile.js
gulp.task('watch', function () {
gulp.watch('src/*.coffee', ['scripts']);
});
Now, running gulp watch
will continuously watch our file system for changes:
$ gulp watch
[ make changes to src/fun.coffee
]
Play around with src/fun.coffee
and see our watch
task respond in the terminal.
// fun.coffee
console.log 'Hello Gulp!'
You can see that gulp
is keeping an open process to re-run our task each time fun.coffee
is changed. AWESOME!
Sadly, we only have one task. It'd be nice to have that be our default
wouldn't it? Running gulp
without arguments would save us a lot of finger fatigue.
Let's update our default
task to run scripts
and watch
when gulp
is run without arguments.
// gulpfile.js
gulp.task('default', ['scripts', 'watch']);
$ gulp
:)
As you can see, Gulp.js is insanely flexible, has very little configuration, and writes like application code.
I think you'll really like using it in your projects.
Who has questions?
Thanks for all, very informative read!