Cost of modules
Find out which of your dependencies is slowing you down
npm install -g cost-of-modules
cost-of-modules in the directory you are working in.
--less Show the biggest 10 modules
--yarn Use yarn instead of npm to install dependencies
--no-install Skip installation
--include-dev Include devDependencies as well - for
Show your support
Help out with windows support!
I recently published a npm module (auto-install) and I wanted to know how many bytes am I making people download before they can use it. Turns out, it was a whopping 30M!
More than space on disk, I want to optimise for install speed - setup is part of the user experience
Now, there are 3 things that you can do to make your npm package smaller
Make sure all your workflow tools are in
devDependenciesand not in
dependenciesThese include your build tools, testing frameworks, etc. Only
dependenciesget installed when someone installs your package. (That being said, for better collaborator experience, you should optimise for both)
Only include the files you need by using
package.jsonor by including a
.npmignore. More on that here.
Use packages which do the job and take the least amount of space. For example, I realised that I did not need
yargs, I only needed their parser
yargs-parserwhich is much smaller.
Bonus: Bundle all your code together and strip out the functions that you don't use - I still have to try this out. This could be the most impactful one.
In my case, the big size was because of # 3, the bunch of npm packages that I was using.
You can't fix what you can't measure
With npm 2.x, it was easy to find how much space is each of your dependencies taking. You could just look at the size of each directory in
With npm 3, the packages are installed in flat manner, so it isn't so straightforward.
That's the reason why I created this tool.
I would want to make this tool a pre-publish checklist.
- See the size of your
- Include the size of your files
- Check if
- Check if there are any
dev dependenciesthat are specified as
- Compare size with the last release. Something like this:
MIT © siddharthkp