when we publish python package, we use pip, which will trigger setup.py to build some necessary packages.
I'm publishing a golang package, I wish when my user gets my package, it will auto trigger a makefile (some scripts to prepare environment) before installation. Is it possible?
I wish when my user get my package, it will auto trigger an makefile (some scirpts) before installation.
That would allow you to execute arbitrary commands on my machine while I do some innocent looking go get github.com/yiw/totallyharmless because I want to use that package.
Is it possible?
I hope not. This happened in the past (through clever cgo stuff) and was fixed as this is a major security issue.
Related
Im developing my first go project and i need to develop it on a remote linux server. Its something small and this is why i need something really simple, but 3rd day now im trying to setup my goland with no success. Starting to wonder if i need it at all here.
I already setup my SSH terminal and my SFTP. The problem is that im not sure how to write my code locally, build it locally and upload to the server or build it on the server. This is what i tried:
Tried to develop with remote file manager and remote terminal, just using goland as editor. The problem here is that if goland meets some missing package, it will give me errors, maybe here i can somehow ignore this?
Tried to write it locally, sync with remote on every save and compile it from the terminal, but this will create an issie with missing packages, but in general it works.
Tried to write it locally, compile it remotely from the Run options. The problem here is that i want to write my install.go file locally and upload it to /root/project/install as compiled file, but its creating some temporary files, tried even to overwrite my -o file from the "Go tool arguments", but it just adds the path to the existent one.
Tried to write it locally and compile it locally and then upload it to the server, but cant find the way for such thing at all.
Maybe dlv, but it looks like a simple issue, hope i wont have to install additional software on the server because of this.
Is some of those options valid or im missing another options? I hope you understood what i really want.
Thank you in advance!
but 3rd day now im trying to setup my goland with no success. Starting to wonder if i need it at all here
You don't need it.
Tried to write it locally and compile it locally and then upload it to the server, but cant find the way for such thing at all.
This is what you want. The only thing you need to install on your server is the executable from go build, built for the server's architecture, which you can copy over SFTP.
When you want to run your program locally, you will use go build to produce an executable that you can run.
When you build for the remote server, you will want to set GOOS and GOARCH to the values appropriate for your server:
GOOS=linux GOARCH=amd64 go build -o my-project.linux-amd64
If your server is ARM, substitute arm64 for amd64.
Then copy my-project.linux-amd64 to the server and it will be able to successfully run there.
Im developing my first go project
Then make sure to understand this point: go executables don't need go libraries at runtime.
This is a big selling point for Go, and is different from an interpreted language like Python, Javascript, Ruby, and also different from languages that run in a software virtual machine like JVM (Java) or BEAM (Erlang) baed VMs.
As a newcomer to go, please make sure to read the tutorial.
Save yourself a lot of hassle and Write tests right away as you develop your first project. These should be your primary way of demonstrating functionality as you work on your project.
A little backstory:
I am creating my first Go web app for school. The professor said that I could use Go. I asked him if I could use the latest version, 1.13, rather than the version installed on the server, 1.10, to leverage the module management feature and the updated errors module. He did not say that I was restricted to 1.10 and even gave me the contact information of the server admin. I reached out to the server admin with my professor CC'd and he said he doesn't want to update the server's version of Go in the middle of the semester. He then included instructions on how to download, install, and use whatever version of Go that I need in my home directory.
Pick your version:
https://golang.org/dl/
Set it up for your own use:
https://golang.org/doc/install
I installed go 1.13 and updated my own envvars to reflect this version and everything works.
Well, the other requirement is that I have to hand my professor my project and that it has to compile on the server. I am thinking that if I hand him a bundle and give him directions on how to build it, then I've technically met the requirements of the project so far.
My question is, does Go have anything that takes my Go v1.13 environment and packages everything up so that the project can compile on the target server? I have only been able to find solutions along the lines of "just copy the project binary to your production server" but that doesn't help me. I need it to compile on the production server. Besides, I tried copying my binary to the production server and it couldn't find my html templates (stored in ./ui/html/ directory) but I guess that will be solved in this discussion or saved for another SO question later.
student#universityserver:~$ ./web
INFO 2019/09/22 10:21:52 Starting server on :5089
INFO 2019/09/22 10:23:03 <ipaddress>:63527 - HTTP/1.1 GET /
ERROR 2019/09/22 10:23:03 handlers.go:29: open ./ui/html/home.page.tmpl: no such file or directory
The only thing I can think of right now is to basically add everything they need (the go amd64 binary distribution and all modules) and write a script that handles it all (extract go in local folder, export envvars, build, etc.)
Dear, Professor.
Copy this tarball to your home directory, extract, run build_my_goapp.sh script.
Sincerely,
Your student
The answer in my case was that I can supply a Makefile that does anything I need it to do.
(It's actually good practice, if not an expectation, of a project that is turned in.)
The requirements were that my project needs to compile on the server and that I had to supply a Makefile. Therefore, any downloading, installing, and setting up of a Go environment for my project can be done, as long as everything happens in the user's environment and is not something that needs sudo to install correctly. The server does not have Docker.
I am curious to know if there are methods where I can add additional packages to the anaconda installer. I am basically looking for a solution for creating an anaconda installer which has some extra python packages added along with it. Thus the participants to whom I give the installer need not be worrying about Internet connectivity or to add additional commands.
This is meant for an introductory hands-on python session. Hence the objective is to make the whole installation process less confusing as possible to the participants of the session.
I am aware of using docker as well as using environments. I am looking for something more simpler, say as seamless as anaconda installation for my participants.
Currently, I am thinking of doing the following.
1) Provide the .tar.gz file of the packages along with the installer
2) After installation and creating environments, install the libraries using pip from the .tar.gz
python -m pip install c:\mymodule\great.tar.gz
Any method which is simpler than the above one is welcome.
From the conda documentation on creating custom channels:
If you do not wish to upload your packages to the Internet, you can
build a custom repository served either through a web server or
locally using a file:// URL.
The instructions on that page tell you how to create a local custom repository from conda packages. They're aimed at people who are building their own packages but as far as I can see you can also use the existing packages that you can download from the repositories at https://repo.continuum.io/pkgs/.
You can then use the file:// URL of that repository in the -c specification of conda create and/or conda install commands to set up the environment for your users to work in.
I'm automating some package installation. We have created the package in question, but let me explain first.
In part of the automation, I'm installing the package with a shell script using dpkg -i.
I'm already using the non-interactive backend, so it's trying to hide the dialog prompt asking for an IP address. However, it doesn't default to an old value and tries to run a MySQL query without the host for the query (yes, I know, bad design; I didn't make the package).
Now, I can edit this package, but that limits the support team with this automation, as they will obviously only be able to patch to this version of the package, and not any previous versions. I'd rather avoid that if possible.
Despite searching, I can't seem to find a way to fill that out, had the thought of something like
dpkg -i some-package.deb <<< "127.0.0.1"
but that doesn't seem to work. Any thoughts on this?
how can I determine the project type before they try to install my package. I wanna show them a cancelation message if the project type is not and MVC application and cancel the installation process?
do I need to add a ps1 script file into tools folder of my package? if so, what should I write inside this script?
Currently it seems that pre-installation checks for aborting the install is not possible. See the answer from David Fowler here.
Scott Hanselman has an example on how to check pre-requisites using the install.ps1 script. He also mentions that it's currently not possible to stop the installation process, but that "In the future there will likely be a pre-install or a dependency check".