Cannot read property 'pid' of null

[kgdev]

·
/home/ubuntu/weidian/backend/node_modules/gulp-develop-server/index.js:188 │············
var pid = gutil.colors.magenta( app.child.pid ); │············
^ │············
TypeError: Cannot read property 'pid' of null │············
at ChildProcess.stopped (/home/ubuntu/weidian/backend/node_modules/gulp-develop-server/index.js:188:45) │············
at ChildProcess.g (events.js:199:16) │············
at ChildProcess.emit (events.js:129:20) │············
at maybeClose (child_process.js:1015:16) │············
at Socket. (child_process.js:1183:11) │············
at Socket.emit (events.js:107:17) │············
at Pipe.close (net.js:485:12)

[kgdev]

var pid = gutil.colors.magenta( app.child.pid );

TypeError: Cannot read property 'pid' of null

[nforysinski]

This issue is most persistent when watching the many results of a js/css set of gulp tasks. For instance, I have several js and css files that need to be minified and linted, which then get moved to a "dist" folder. My develop task is as follows:

gulp.task('watch', ['server:start'], function () {

    var notify = function(file) {
        server.changed( function( error ) {
            if( ! error ) lr.changed( file.path );
        });
    };

    gulp.watch(paths.css, ['sass']);
    gulp.watch(paths.html, ['compresshtml']);
    gulp.watch(paths.js, ['lint', 'compressjs']);
    gulp.watch(['./public/dist/**/*']).on('change', notify);
});

This yields the child pid error as well as several other event errors.

[chimerast]

I think this issue is caused by a bug of checking multiple restarts.

https://github.com/narirou/gulp-develop-server/blob/master/index.js#L206-L210

    // already called this function
    if( isChanged ) {
        isChanged = false; // *** I think this line cause the problem. It is unnecessary code. ***
        return done( null, 'Development server already received restart requests.', callback );
    }

It calls the kill function multiple times when the watcher task calls restart multiple times in a very short time.

[narirou]

@kgdev, @nforysinski, @chimerast
Thank you for the detailed report.

As it is said, the value of isChanged was incorrect.
This bug made it possible to send a restart request many times.

But, as a result, I do not know whether this issue has been fixed yet.
I plan to verify the reproducibility of this issue, and add more test code.

[narirou]

I think this error fixed by previous commit. (thanks @chimerast)
If not, please reopen this issue. Thanks.