In *nix, when I compile software, when should I do:
# make install
vs
# make install clean
?
When you call make with arguments, you are giving it a list of targets.
There is some discussion of this in the Upgrading Ports section of the FreeBSD Handbook where it says:
Unlike the usual make install clean command, it will upgrade all
the dependencies prior to building and installing the selected
port.
[...]
Note: You can save two extra steps by just running make install
clean instead of make, make install and make clean as three
separate steps.
Also, as a side note: Don't build your package as root unless you really need to. In general you should work in an unprivileged account and then as a final step you can do sudo make install if you have to. It would be better, though, to add the unprivileged user to a group with write-access to the install directories, or even better to install it into a non-system area (for example, with ./configure you can use --prefix=) since most systems, whether Linux or BSD, usually have packages and a packaging system that is used to install software.
Related
I am trying to change the version of nodejs using cmd n installed globally by yarn. Neither yarn is detected by sudo nor n.
Please help me to fix this:
When running the n command to change node version, it shows permission is denied by many files to be written by n
Two approaches to avoid permission issues without requiring use of sudo when running n:
install and run Node.js from your home folder using N_PREFIX and PATH (see also n-install which automates this)
if it is your personal computer, take ownership of the folders used with default install location of /usr/local
See the project README for more: https://github.com/tj/n
(Disclaimer: I am the maintainer of n.)
I took a Vue 2 online course and it did show (but didn't really explain) how to install node.js, npm and vue. Currently using vue-cli to set up my project using vue init webpack-simple. Problem is I have a Windows desktop and a Mac laptop. I'm using Box cloud on both but I need to have 2 separate folders for the same project. Basically, project-1-windows and project-1-mac.
I can't run npm run dev on the project-1-mac while on Windows 10 and vice versa. The only way I know to run both is to delete the node_modules folder and run npm install. However it takes a while for the files to download. Is there an easier way to do this?
It looks like you want either GitHub/BitBucket/friends or (for much more complex set-ups) Docker.
I will explain only the first (easier) option. To set-up docker, you rather go to its docks.
So, for GitHub/BitBucket/friends way, you need some one-time set-up (you have to do all of this in terminal of your machine. I don't go too deep into details because you may find corresponding docks for each thing easily by googling it).
Install git if needed on both machines. On mac, you just run git --version in terminal. It'll either show the version of installed git or will ask if you want to install git together with other developer tools.
Install brew on Mac, install any of these on Windows. These are just package managers. Use them to install nvm.
Install nvm (it's node version manager, arguably the most convenient way to manage node.js installations) on both machines.
Use nvm to install node.js (npm comes bundled with it) on both machines. That's it! One-time set-up done. Run node -v && npm -v to check that both are installed.
Now, to start each project you would do the following:
Create a repo (which is like a folder but on GitHub/BitBucket server) that you may freely access from any device that has internet connection.
Start project on any of your machines with something like npm init or vue init webpack-simple or whatever you feel comfortable with.
Run git init
When you do changes, commit & push them into your online repo.
Avoid committing files that might be auto-regenerated, they simply don't worth storing.
You may use any npm commands.
When you want to continue working on another machine, simply git clone your existing repo, run npm install and you are done.
Commit changes if needed.
git pull changes to another device if needed
I'm getting errors with npm while trying to install/update packages without SU permissions on Linux.
The easy way to solve the problem is execute sudo npm install <package>, but I'm not sure if it is a good idea.
Best way is to become the owner of .npm folder, as I found into StackOverflow's questions and blog posts.
My question is: why run npm as SU it isn't a good idea?
Running npm as a super user has a risk of running some untrusted code as a super user which can potentially mess with your entire system. Running npm as an unprivileged user has a risk of running that code with less privileges and it won't be able to mess with the entire system - just with your own files (which can be equally bad, depending on how you look at it).
What I often do and recommend is to install Node in your home directory instead of globally on the system if it's your own computer. That way you don't have to run with sudo or su for npm or even for make install of Node itself.
I run a lot of versions of Node that I compile from sources sometimes with different switches and the convention that I use is to install Node in versioned directories, either globally in /opt (but then you need sudo) or locally in my home directory in ~/opt.
I do it like this:
wget https://nodejs.org/dist/v7.1.0/node-v7.1.0.tar.gz
tar xzvf node-v7.1.0.tar.gz
cd node-v7.1.0
./configure --prefix=$HOME/opt/node-v7.1.0
make && make test && make install
Then I create a symlink ~/opt/node pointing to ~/opt/node-v7.1.0 and I have:
PATH="$HOME/opt/node/bin:$PATH"
in my .profile or .bashrc.
That way I don't have to run as super user for installing Node or for running npm.
As a bonus I can quickly switch my default Node version just by changing the symlink, and at any time I can run any other version if I change the PATH or run Node with a full path like ~/opt/node-v7.0.0/bin/node.
I explained that installation process in more detail in my other answers:
node 5.5.0 already installed but node -v fetches with “v4.2.1” on OS X & homebrew?
NodeJS API with external deps in other language
I don't want to go into too much detail here since this answer is about why running npm as a superuser is not a good idea - this installation process is just one solution to not have to run npm as a superuser.
Other options of setting your npm permissions to avoid running as a superuser are described in Fixing npm permissions in npm docs (thanks to RyanZim for pointing it out in the comments).
I want to install nodejs on my PCBSD 10 system. I have downloaded the src file of latest nodejs. On terminal I ran the command ./configure It runs fine. After that I tried make but its asking on console
I need GNU make. Please run 'gmake' instead.
Then I tried gmake the terminal says
CORRECT>gmake (y | n | e | a)?
I pressed y then again it says "I need GNU make. Please run gmake instead"
How to install nodejs???
If you install it using the FreeBSD ports system, that will take care of these details for you.
If you haven't installed the ports tree, run the following command as root;
svnlite checkout https://svn0.us-west.FreeBSD.org/ports/head /usr/ports
Note that there are some subversion mirrors for FreeBSD. Choose the one closest to you. After the initial checkout, you can update the ports tree by running the following command as root;
svnlite update /usr/ports |& less
When the ports tree is installed and up-to-date, you can install node by running the following commands as root:
cd /usr/ports/www/node
make install clean
This will first install the ports that node depends on (gmake and python2) and then node itself.
You might also want to install the node package manager from the www/npm port.
I setup a development environment with Windows 8 and Ubuntu as a virtual machine. For that I use VirtualBox.
I also manage to create a shared folder in VirtualBox.
In this shared folder I try to start a project with ember-generator of Yeoman.
yo ember --skip-install --karma
npm install --no-bin-links
For installing modules NPM I use the option "--no-bin-links" not to create symbolic links. Unfortunately, I still have errors creations symbolic links ... Is what I use although this option ? There he has a bug ?
The NPM docs about parameter "--no-bin-links" say:
will prevent npm from creating symlinks for any binaries the package
might contain.
Which will just cause NPM to not create links in the node_modules/.bin folder. I also searched for a way to prevent NPM from creating symlinks when using npm install ../myPackage, but can't find any solution...
Update: The npm support team said this will reproduce the old behaviour (no symbolic links):
npm install $(npm pack <folder> | tail -1)
Works for me in git-bash on Windows 10.
This Stack Overflow page comes up in Google search results when trying to solve the issue of installing local modules (ie. npm install ../myPackage) and not wanting symbolic links. So I'm adding this answer below to help others who end up here.
Solution #1 - For development environment.
Using the solution proposed by the NPM support team as mentioned in the other answer works...
# Reproduces the old behavior of hard copies and not symlinks
npm install $(npm pack <folder> | tail -1)
This is fine in the development environment for manual installs.
Solution #2 - For build environment.
However, in our case, the development environment doesn't quite matter as much though because when committing our changes to Git, the ./node_modules/ folder is ignored anyway.
The files ./package.json and ./package-lock.json is what is important and is carried into our build environment.
In our build environment (part of our automated CI/CD pipeline), the automation just runs the npm install command and builds from the dependencies listed in the package.json file.
So, here is where the problem affects us. The locally referenced files in the dependencies list of the package.json causes symlinks to appear. Now we are back to the old problem. These symlinks then get carried into the build's output which move onto the Stage and Production environments.
What we did instead is use rsync in archive mode with the --copy-links option that turns symbolic links into copies of the original.
Here is what the command looks like in the automated build:
# Install dependencies based on ./package.json
npm install
# Make a copy that changes symlinks to hard copies
rsync --archive --verbose --copy-links ./node_modules/ ./node_modules_cp/
# Remove and replace
rm -r ./node_modules/
mv ./node_modules_cp/ ./node_modules/
I have a similar environment. Apparently the Virtualbox (vagrant) synchronisation has problems when renaming or moving files, which happens when updating modules.
If you do a file listing (ls -alhp) on the command line and see ??? for the file permissions, then it is time to reboot your virtualbox. This will set the permissions to valid values. Then use the --no-bin-links option when installing a module.