Ubuntu 20.04 - is it possible to install the Linux headers package as non-root? Perhaps to a different directory under $HOME somewhere?
The reason for asking is I would like to have an Jenkins job that periodically fetches and, if necessary, installs the latest version before building a driver against those headers. The Jenkins job will run as a non-root user, so I don't really want to mess with passwords or keys if I can do the whole thing with 'normal' user permissions.
(I have thought about using DKMS, but the machines where the driver will be installed are very restricted and we can't install build tools on them.)
Thanks.
Related
I have installed Anaconda recently in a new Windows computer. I have no experience with managing installed packages in Windows, but in Linux. I created a new project with PyCharm and chose to use as interpreter Python 3.7 that I installed separate from the Python 3.6 version that Anaconda comes with. Now I want to be able to use Anaconda modules such as numpy or pandas in my PyCharm Project, that is using a virtual environment set up in a PyCharmProjects folder.
I fail to understand how exactly should I do it. What is that I have to copy or run to migrate the Python packages installed in Conda to my Virtual environment created from a clean independent install of Python 3.7? When I try to import them they don't work. Do I have to create a new project and migrate the files or can I do it without that?
Edit
Also, I am very unclear about how I can run Python36 that Anaconda installed. There is no conda command in Windows and python runs python27. How do I do this?
So it seems that using Python in Windows goes a bit differently than in Linux/MacOS. In order to change the default version that runs when python command is executed, one has to run the command regedit and Find (Ctrl+F) Python.exe, and change the path to default Python interpreter that one wants to use. Python installed by Anaconda can be found in Anaconda's folder in C:\ProgramData (if installation was for all users, if not its probably in some other folder in C:\Users\[User].
As for the other part of the question, I'm not sure but I think packages have to be re-installed in the virtual environment, unless you know how to copy the files one by one. Also, it's important to be careful with the version compatibility. In my case, the two Python installations are different versions, so it might not work the trick to copy the files. Other option is to change where the interpreter runs; if in that folder the packages are installed, the installation will succeed. You can also change Python's path to find packages; but that is something that must be done programatically and is not very handy to have to be running those lines of code each time.
When creating a project (or even when the project is already created) you can choose to change the interpreter to Anaconda's Python, even if you are not using conda as a package manager, but virtualenv by Python. That'd be the trick for me. Aditionally, PyCharm also natively integrates package installation into the virtual environment in a GUI menu.
If anyone has a better answer/explanation, I'm looking forward to getting to know it, but in the meanwhile that is the conclusion I have arrived to.
I have two CentOS VMs which use Jenkins to run automated tests through firefox. Both have firefox installed. Both versions of firefox are the same (firefox-56.0.1). I do not know if they are both x86_64 or some other type, but whichever they are, they are the same. (I am using the same .tar.bz2 file. I copied it from one instance to the other.)
In one instance, I am able to run firefox. "firefox --headless" returns "*** You are running in headless mode."
After copying the .tar.bz2 file to the other instance and installing firefox, I find that the new instance does not have the same performance. "firefox --headless" returns the following:
XPCOMGlueLoad error for file /usr/local/firefox/libmozgtk.so:
libgtk-3.so.0: cannot open shared object file: No such file or directory
Couldn't load XPCOM.
What might explain this difference? It appears that I did something right in the first instance, but I cannot tell what. Is there some setting that will prevent firefox from trying to use the "libgtk-3.so.0" file? This file does not exist on either instance.
I've seen this question elsewhere, but the answer seems to focus on versions (I know my version has worked in the first instance) and a particular bug that has been backlogged (this bug only bothers me for the new instance).
Problem solved minutes after asking the question.
Simply run:
yum info gtk3
Check if gtk3 is installed. If not:
yum install gtk3
My old instance had gtk3 version 3.22.10 installed. My new instance had the same available but not installed. I don't recall seeing this in any of the guides to running firefox headlessly, but a search result that I did not originally think worth checking was able to resolve this rather quickly. Credit to the folks at https://forums.fedoraforum.org/showthread.php?310652-Udating-Firefox-fedora-19
For Ubuntu or Debian distros use:
apt-get install packagekit-gtk3-module
You probably installed the wrong firefox version for 32bit on a 64bit system.
try downloading firefox with this portable installer download-mozilla-portable.sh which should work on a 64bit linux (tested in Ubuntu 19.10)
I am using MacOS and vagrant for Go development. I don't want install any development packages like Go, npm etc. on my host machine (MacOS). All of dev softwares installed on vagrant.
But there is a problem with vscode-go plugin, Because it requires few go packages like go vet, goimport etc. those packages installed on remote host (vagrat)
So how can I say to vscode that run these commands on remote host?
Make a nfs share on the GOPATH environment variable in vagrant. On your Mac define also a GOPATH env variable. That way vscode can install the tools on your vagrant.
Although not sure this covers Mac vs Linux binaries.
Another approach might be too create bash aliases/functions which invoke the command over ssh.
I wrote a blog post on how to set up vscode on windows. I would really recommend to install tools natively. Go tooling is small, super fast and doesn't add lot of bloat to your system. E.g. PHP ads lots of crap etc.
https://marcofranssen.nl/start-on-your-first-golang-project/
Might clarify some stuff for you.
when installing MSysgit, also MSys (as the name suggests) is installed on my machine.
When I additionally install DevKit for Ruby, yet another copy of MSys is installed on my computer. When installing the Haskell Platform for Windows, again another copy of MSys is installed on my machine. Is there any way of installing MSys/mingw once and tell all other software to do a lookup in the Path?
The version of msys that is included in the Git for Windows package is modified and if you attempt to replace it with a stock msys package you will run into problems. I can't comment on the other packages but basically - it is not worth worrying about. Disk space is significantly cheaper than the time you spend trying to make a number of independent packages share a common msys platform.
However, there are in fact people working on trying to sort that out. msys now has a package management system and I know of at least one project attempting to get the Git for Windows build environment working withing that system mingwGitDevEnv.
A project I am working on requires an old version of SLAPD (the LDAP
server) and must run on Windows, hence I am using cygwin packages:
I've found a binary package of slapd version I need 2.2.x here:
http://www.mirrorservice.org/sites/sourceware.org/pub/cygwin/release/...
How do I manually install it in cygwin? It does not seem to be a package-manager file format (like an RPM or a Deb), but is there a tool to install it for me?
FYI, I'm operating behind a corporate firewall that prevents me from viewing the main cygwin site (including the documentation), so please nobody tell me to RTFM - I wish I could!
Thanks
If you need an older version, the Cygwin time machine might help. To use it, I had to invoke setup.exe from the command line, giving it the --no-verify/-X option and it still wouldn't let me downgrade the "cygwin" package itself but at least it is, albeit very slowly, allowing me to install the version I wanted to test as a parallel installation.
If you are using the cygwin package installer to install this package, usually if you select to view the full info and scroll to the package you need you will see some info about the version.
Now if you click on the version number you will see that it changes from the oldest in the repository to the current. If your repository has the version you need then just finish the wizard.
Some cygwin mirrors keep old package files, for example, http://mirror.isoc.org.il/pub/cygwin/
You can use Sonatype nexus to mirror an existing cygwin repository, but provide your own setup.ini .
Copy the already downloaded repo content to c:\progs\nexus\sonatype-work\nexus\storage\mirror.isoc.org.il-pub-cygwin\
Put your own setup.ini and setup.bz2 there.
Start setup.exe with --no-verify
Paste
http://localhost:8081/nexus/content/repositories/mirror.isoc.org.il-pub-cygwin/
into the mirror selection box and click "Add"