Im trying to use gulp and jscs to prevent code smell. I also want to use watch so that this happens when ever a change is made. The problem I'm running into is jscs is modify the source file that is being watched. This causes gulp to go into an infinite loop of jscs modifying the file and then watch seeing the change and firing off jscs again and again and again ...
const gulp = require('gulp');
gulp.task('lint', function() {
return gulp.src('/src/**/*.js')
.pipe(jscs({
fix: true
}))
.pipe(jscs.reporter())
.pipe(gulp.dest('/src'));
});
gulp.task('watch', function() {
gulp.watch('/src/**/*.js', ['lint']);
});
It's generally a bad idea to override source files from a gulp task. Any Editors/IDEs where those files are open might or might not handle that gracefully. It's generally better to write the files into a separate dist folder.
That being said here's two possible solutions:
Solution 1
You need to stop the gulp-jscs plugin from running a second time and writing the files again, thus preventing the infinite loop you're running into. To achieve this all you have to do is add gulp-cached to your lint task:
var cache = require('gulp-cached');
gulp.task('lint', function() {
return gulp.src('/src/**/*.js')
.pipe(cache('lint'))
.pipe(jscs({
fix: true
}))
.pipe(cache('lint'))
.pipe(jscs.reporter())
.pipe(gulp.dest('/src'));
});
The first cache() makes sure that only files on disk that have changed since the last invocation of lint are passed through. The second cache() makes sure that only files that have actually been fixed by jscs() are written to disk in the first place.
The downside of this solution is that the lint task is still being executed twice. This isn't a big deal since during the second run the files aren't actually being linted. gulp-cache prevents that from happening. But if you absolutely want to make sure that lint is run only once there's another way.
Solution 2
First you should use the gulp-watch plugin instead of the built-in gulp.watch() (that's because it uses the superior chokidar library instead of gaze).
Then you can write yourself a simple pausableWatch() function and use that in your watch task:
var watch = require('gulp-watch');
function pausableWatch(watchedFiles, tasks) {
var watcher = watch(watchedFiles, function() {
watcher.close();
gulp.start(tasks, function() {
pausableWatch(watchedFiles, tasks);
});
});
}
gulp.task('watch', function() {
pausableWatch('/src/**/*.js', ['lint']);
});
In the above the watcher is stopped before the lint task starts. Any .js files written during the lint task will therefore not trigger the watcher. After the lint task has finished, the watcher is started up again.
The downside of this solution is that if you save a .js file while the lint task is being executed that change will not be picked up by the watcher (since it has been stopped). You have to save the .js file after the lint task has finished (when the watcher has been started again).
Related
I am working on a simple gulpfile and noticed an issue with gulp.watch method. If I add a new file to an empty directory gulp.watch will not fire. However if there is at least one file in the directory all change events are detected. I could obviously restart my "watch" task every time there is an empty directory added with a new file or I add a file to an existing empty directory but that seems counter intuitive to the purpose of gulp.watch method.
To be clear watch does detect files that are added and deleted only after at least one file exists in that directory.
My question is wether or not this is a bug exclusive to me or if more people have experienced this. Also does anyone know of a current work around?
Here is my gulp task:
gulp.task('watch', () => {
var watcher = gulp.watch('src/styles/scss/*.scss', {cwd: './'}, ['styles']);
watcher.on('change', (event) => {
console.log(`File ${event.path} was ${event.type}, running tasks...`);
});
Current gulp version: 3.9.1
P.S. I also know this may be a limitation of the technology I just don't what to report a bug to the gulp team that isn't a bug.
Thanks!
Awesome! Thank you, Mark for getting me in the right direction. It is not a bug there is just a specific way you have to do it.
gulp.task('watch', () => {
var watcher = gulp.watch(['src/styles/scss/*.scss', 'src/styles/*],{cwd: './'}, ['styles']);
watcher.on('change', (event) => {
console.log(`File ${event.path} was ${event.type}, running tasks...`);
});
The trick is watching your parent directory for any changes. This will now detect file changes as well as added and deleted files in empty subdirectories.
So I have a task like so:
gulp.task('scripts', function() {
return gulp.src(['app/scripts/app.js', 'app/scripts/controllers/**/*.js', 'app/scripts/services/**/*.js', 'app/scripts/directives/**/*.js', 'app/scripts/libs/**/*.js' ])
.pipe(concat('external.min.js'))
.pipe(ngAnnotate())
.pipe(uglify())
.pipe(gulp.dest('app/scripts'))
.pipe(gulp.dest('dist/scripts'))
});
and I have a watch task:
gulp.task('watch', ['sass-dev', 'scripts'], function() {
gulp.watch('app/style/sass/**/*.scss', ['sass-dev']);
gulp.watch('app/scripts/**/*.js', ['scripts']);
});
All works well, except whenever I save a JS file, "scripts" runs multiple times. I'm assuming the problem lies with the gulp.src and it looking at multiple files, but I'm not sure.
This is no big deal (to me), but sometimes I'll swap over to the command line and the task is running infinitely. It just keeps getting called over and over again.
If you haven't already guessed, I'm running Angular, which is why app.js is first and I have ngAnnotate.
Can someone shed some light on why the script runs continuously sometimes?
I guess the problem is .pipe(gulp.dest('app/scripts')). You're doing some stuff (uglify and angular stuff) with your scripts and then you place them in the same folder you're watching. So the scripts task will launch again and again and again.
You should remove this line and only place your distribution scripts in your distribution folder and leave your app files untouched.
I'm using gulp to convert markdown files to HTML, and using the gulp-watch plugin (not the gulp.watch API function) to rebuild files if they change. Works great!
gulp.task('markdown', function () {
gulp.src('src/**/*.md')
.pipe(watch('src/**/*.md'))
.pipe(markdown())
.pipe(template('templates/md.tpl'))
.pipe(gulp.dest('dist'));
});
The problem is that the pipeline src is the markdown files, but within the pipeline I also reference a template file. If that template changes, all the markdown files need to be rebuilt. Is there a way to express that dependency in gulp/gulp-watch?
I tried using gulp.watch (the API function) to watch the template and run the 'markdown' task if it changes ...
gulp.watch('templates/md.tpl', ['markdown']);
... but that didn't work. Nothing happens. I assume having gulp-watch in the pipeline prevents it from doing anything.
I guess I could create a two tasks, one with gulp-watch and one without, and use the one without to force a full rebuild. I'd rather not, because then it becomes an ongoing problem to keep the two in sync.
Is there a better way?
I guess I could create a two tasks, one with gulp-watch and one without, and use the one without to force a full rebuild. I'd rather not, because then it becomes an ongoing problem to keep the two in sync.
Remember, gulp is just JavaScript.
Simply write a function that constructs the stream with or without the watch() step depending on a parameter that you pass. The gulp-if plugin let's you write something like this in a very concise way (although it's not necessary and could be done without it).
Here's how I would do it:
var gulpIf = require('gulp-if');
function processMarkdown(opts) {
gulp.src('src/**/*.md')
.pipe(gulpIf(opts.watch, watch('src/**/*.md')))
.pipe(markdown())
.pipe(template('templates/md.tpl'))
.pipe(gulp.dest('dist'));
}
gulp.task('markdown', function() {
processMarkdown({watch: true});
watch('templates/md.tpl', function() {
processMarkdown({watch: false});
});
});
You can specify gulp src as an array, too:
gulp.src(['src/**/*.md', 'templates/md.tpl'])
I have the following code fragment in my gulpfile.
gulp.task('static', function() {
return gulp.src(['./src/**', '!./src/js/**', '!./src/js/', '!./src/scss/', '!./src/scss/**'])
.pipe(gulp.dest(outputDir + '/'))
});
gulp.task('watch', function() {
gulp.watch(['./src/**', '!./src/js/**', '!./src/js/', '!./src/scss/', '!./src/scss/**'], ['static']);
});
gulp.task('dev', ['static']);
gulp.task('default', ['watch', 'dev']);
If I run gulp dev, gulp watch or gulp static, everything works fine. However, if I run just gulp (default), it does the static task 5 times. Can anyone help me out with why this is happening?
P.S. The paths passed to watch are such because if I don't disclude the directories as separate paths, it seems to be copying the empty directories js and scss for some reason.
Probably because you're not returning the tasks, and you need them to be asyc.
See this: Gulp.js task, return on src?
and the docs (also linked in SO post above) https://github.com/gulpjs/gulp/blob/master/docs/API.md#async-task-support
Also, the dev task looks redundant in its current form - you may as well use the task static directly, unless you plan to bundle in more tasks with dev
I'm using gulp-watch plugin and would like to copy newly added files in the source to the target destination.
watch({glob:SOURCE + '/**/*.js'})
.pipe(plumber())
.pipe(gulp.dest(DESTINATION));
Every time a new file is added into the SOURCE directory I get "Bus error: 10" and the watch breaks without copying the newly added file.
Please you this syntax for adding new files in gulp
gulp.task('task_name', function() {
return watch({
glob: SOURCE
}, function(files) {
return files.pipe(plumber()).pipe(jade()).pipe(gulp.dest(DESTINATION));
});
});
gulp.watch doesn't create a source stream, it triggers on file changes and calls tasks.
Try creating a simple move task alongside a watch task, then trigger the move from the watch. Something like this:
gulp.task('move-js', function() {
gulp.src('./js-src/**/*.js')
.pipe(gulp.dest('./js-dest'));
});
gulp.task('watch-js', ['move-js'], function() {
gulp.watch('./js-src/**/*.js', ['move-js']);
});
Note that the watch-js task has move-js as a dependency, this will call the move task whenever the watch is invoked, rather than waiting for something in the watched directory to change.
I repeated the glob for clarity, but that should probably be stored in a variable.