Is there a best practice or common task(s) that people use to separate their source code and minified code? I don't want to have to re-minify and concatenate a file manually every time I edit it in the src/ directory.
Apologies if this is an extremely obvious answer, but I cannot seem to find any good Gulp tutorials on folder structure dependency.
Thanks in advance.
One common practice is to have a dev task that runs gulp-watch when you're working on something. This task (probably) shouldn't minify or run any other production type tasks. Then when you're ready to deploy you run your prod task which will minify and add cache busting, whatever you need (without gulp-watch).
In terms of folder structure you can have a /src and /dest directory if you'd like. It keeps things clean in your src directory and you can clean/delete the entire contents of the /dest directory because you know those are all generated files.
Related
In my project I use Webpack mainly for bundling .js and .css files.
Main question is about images. When images are used in .css files Webpack process them and exports to /dist folder. Which is fine, and works like a charm.
What I want to accomplish is pretty same story but with .html files. But! Html files are in different location then my wepack-app.
-/root
--/design
-----/src
--------/js
--------/css
--------/images
--------/...
-----/dist
--/templates
--/...
Is it possible to e.g pass additional path to scan for assets?
I don't want to produce new html. Just check which assets are used in html files from root/design/src/images then process them (same as from css files) and copy to /dist.
You can add a new entry point (a js file) that will require all the html files that you want to be processed.
You will need to install html-loader in order to allow webpack to "understand" html files.
I'm working on a script that should be able to add additional information to a .chm file.
After decompiling it with hh.exe -decompile outputFolder fileName.chm command, I get the html files, and other 2 files with .hhc and .hhk extension.
After editing the html files, I'd like to recompile the files into a single .chm file. I read that that I also need a .hhp file in order to do that, but that's not generated in the decompilation process.
How can I solve this?
This is a problem of Compiled Help Modules (CHM). And yes - you need a *.hhp for compiling again by HTMLHelp Workshop or e.g. FAR HTML.
You know, you can use 7Zip or just open a command prompt window on a Windows PC and type the following:
hh.exe -decompile <target_directory> <path>\<filename>.chm
The only decompiler with any additional features is KeyTools as this can try to rebuild the project (.hhp) file. You'll need this file if you want to recompile the help project.
One thing to note is that the decompile/recompile process isn't a "round-trip" process. Certain features that the help author added to the original help file can't be recovered when you decompile it, so these may no longer work properly after you've recompiled.
This is especially true in the area of context-sensitive help, which may be broken in the new version of the file.
It can be useful, to include the .hhp file itself - after regenerating is done - into the section [FILES] of the project file (.HHP). Thus, this is included in the Compiled Help Module (CHM) when compiling. The appropriate *.HHP file then is decompiled in addition to the other files for future use.
I have a pre-existing project that is currently using gulp.
The key libraries/frameworks/languages are:
MongoDB - Mongoose
AngularJS - With ui-router, also using ngClassify
ExpressJS - With Passport
NodeJS
Jade
Coffeescript
Sass - '.sass' format
JPG/PNG's
Currently everything is watched using live reload, minified using uglify and gzipped. My angular html view/directive snippets are sent into a template cache js file. Even the images are minified using image min.
The single page app is very modular by design, there are multiple 'pages' to the app, each page has a specific use (Take the profile page for example), using ui-route to nest views. Not all users will use each page. Hence why I am choosing to move towards WebPack with each 'page' being a module. The goal for this application is to be as reactive as possible. With potential mild load times when switching which page/module they are on.
My current project structure has a src and dist directory each with a server and client folder. the list directory can of course be safely deleted with every build. I currently have no raw js files or raw html (aside from the gulpfile.js that just requires my gulpfile.coffee), everything gets preprocessed by gulp and thats it.
So here are my questions:
Do I replace most of my gulpfile with webpack, and let webpack process everything (Whats the advantage of this). Or do I create an intermediary folder (The gulp output), then run webpack on that folder (just dealing with the minified js/css/html files). Basically, knowing what my libraries/frameworks are, and my situation, how would you structure the build process?
Can you use an ngClassify app.coffee file as an entry point? Or does it have to be compiled first. (If you can, how?)
You can certainly use Gulp to trigger your Webpack build and manage other tasks you may have however the idea of Webpack is that it is your entire build, you no longer need Gulp tasks to 'minify, 'concatenate' and 'imagemin' files etc as Webpack does all this for you by using Plugins and Loaders.
You will have to run Webpack on the project source, not an already minified bundle created by your custom Gulp build.
The angular questions I don't have an answer to I'm afraid :)
I just want to get an opinion on my workflow. I am aware of Yeoman and have on purpose decided not to use it. My workflow goes like this:
Run bower install to install all project assets dependencies.
Run grunt which copies all js files from the bower components folder to a new js folder and all css files to a new css folder.
Further use grunt task to concatenate and minify all js and css files from the new folders and put them in a dist folder.
Refer to the final minified css and js in dist folder from HTML.
One thing i certainly don't want to do in my grunt task is to perform dependency specific task e.g. grab all js file from bootstrap folder into the new js folder, then grab all js file from prettyphoto folder into the new js folder. I want the grunt task to be as generic as possible so that i can use the same gruntfile in any project no matter what the bower dependencies might look like. The reason is if i should spend all those time writing my gruntfile for each project, why would i not just grab the source codes for all the dependencies in conventional way.
So there is a grunt-contrib-copy plugin to copy files from one place to another which i use to grab all js files from inside the bower's components folder. The problem is most of the bower components come with regular js and minified version of it. So, i am copying both of them and concatenating and uglifying them. So duplicate code!
Does my workflow makes sense? Is so, how can I get rid of the problem I mentioned in the paragraph above?
If I'm understanding correctly, you should take a look at grunt-usemin. You can wrap your js tags in <!-- build:js js/foo.js -->. The useminPrepare task that's included in the package will cycle through any scripts (or css, or images, etc.) that are there and dynamically add them to the concat or uglify task.
The one downside I've found is that the usemin task is fairly slow but hopefully if this pull request is implemented, things will get much, much faster.
When I use yeoman init angular:allto bootstrap my app, yeoman uses the name of the directory I'm calling it from as the name of the main application .js file. I.e. if I'm running the command in /Users/myusername/dev/projects/my-angular-app, then the file app/scripts/my-angular-app.js is created as the main .js file for the app.
If I simply rename the root directory, then using Yeoman to add routes won't work, as it can no longer find the .js file it created. I've tried adding the name of the app as a parameter when invoking the Yeoman init command, but that doesn't work.
I've tried doing the init command in a directory with the name I want the app to have, and then renaming the directory afterwards. But it seems Yeoman reads the name of that directory every time it's invoked, so after changing the name of the directory, Yeoman is no longer able to create routes for me as it can't find the .js file with the same name as the root directory.
How can I make Yeoman use a different name for the my-angular-app.js file?
There's an open ticket about it, please weigh in your thoughts there.
I am not sure if this is possible but you might want to have a look at the Gruntfile.js and replace all the instances in the config were it is storing the old path and replace it with the new path. I am not sure if that will work as I have never tried but it might based on my understanding of Grunt.
I am like you, I like to have things named the way I want which why I don't use Yeoman even though I would love to (like they say, it is opinionated, just too opinionated for me), I have a few custom built grunt tasks that give me most of the functionality I need (mainly auto compile of SASS/TypeScript, minifying CSS/JS, and concating CSS/JS files).
The issue is with generator-angular.
Currently, the app name is taken from the bower component.json file. If you are using the more recent convention of naming this file bower.json, the app name will fall back to the directory name.
Rename bower.json to component.json until support for the new naming scheme makes it into generator-angular.
It looks like these changes have already been made but have yet to be merged into the master branch: renaming-deprecated-component_json