I have a very no-exceptions rule about encouraging people to do a curl|bash install and would just remove that. Provide a link to the script, people can run it if they want. Encouraging the behavior of just directly running scripts off the internet is a bad habit.
In your Proxmox console, enter the following command: bash -c "$(curl -fsSL https://raw.githubusercontent.com/…)
Do not do this. Never run scripts like this directly without inspecting them first. Do not tell people to run your exciting new script like this. Provide a link to the script and encourage users to inspect it first then run it.
Read the comments. Self hosters are little more than users anyway.
Yeah - it’s remarkable that I receive pushback about it. I guess it’s down to the technical immaturity of your average home-gamer vs. people who support Linux systems for a living?
Why would any sensible competent tech user copy paste from other places because this one worked.
Because sites like this and people like you are normalizing the practice. I have seen numerous curl | sh commands pasted on lemmy telling people “how easy it is to install blank”.
Upgrade what? The LXC/VM you just removed because of a wonky script?
Did you purposefully misunderstand me? How did you not know that I meant “how do you update the thing you installed with a rando shell script” and not “how do you update something after removing it”?
You can install with package managers and include with it a helper script to setup the service. No big deal.
But can you spot the difference between http://myservice.com/script.sh and http://myserv1ce.com/script.sh if you use a font that doesn’t make it clear? If you get people used to just copy/pasting/running scripts then there’s a risk they’ll run something entirely different by accident.
There’s no good reason to install things this way.
Neat. Now you have a snowflake install. How do you upgrade it?
The URL can point to a different file. People can post maliciously similar URLs and trick you into running something else.
With a repository you have some semblance of “people have looked at this before”. Packages are signed and it will provide a standard way to uninstall and upgrade in the future.
There’s literally no good reason to replace it with a shell script on a website.
The step I want to skip is having to go back and tab over lines when I’ve moved code or when things don’t paste properly and tabbing gets messed up.
Imagine if there were languages that let you just explicitly state where code blocks start/end and let the IDE handle the formatting for you?
I deal with the syntax on a daily basis though. Moving code around is objectively more difficult with Python since IDEs can’t always know where the block should be tabbed to. Sometimes it is correct at guessing, sometimes not. And when it’s not all I can think of is “why the fuck am I dealing with this?”. It’s a non-issue in proper languages. Just simply doesn’t exist. The IDE there can know where things should be tabbed to. It’s a problem created explicitly by the language syntax.
The best argument I’ve heard for whitespace blocking is “it’s not that bad when you get your text editor configured”. That’s an excuse, not a reason.
Package managers only install a package with defaults. These helper scripts are designed to take the user through a final config that isn’t provided by the package defaults.
This is trivially solved by having a “setup” script that is also installed by the package manager.
That said, let’s not gate keep.
This shouldn’t be an excuse for promoting risky behavior.
With OOP languages, class methods can create a dichotomy of expected returns: Using a list reversal as an example, will a list.reverse(some_list) function reverse the list in-place, or return a copy of the reversed list and leave the original in-tact?
Mmm - that is an excellent example, though not one specific to OOP… But I see your point. I think Kotlin and other languages have addressed these types of issues with ‘mutability’ as part of the type where it becomes explicit which variables can be modified. Kotlin even has “MutableList” as distinct from “List” which lets you say whether a list’s contents can be modified.
I’m not speaking for everyone. Just everyone with taste.
Never a problem.
It solved a problem that didn’t exist and created problems that hadn’t previously existed.
There’s a reason every python “intro” begins with “spend 20 minutes setting up an editor to deal with whitespace” properly.
It makes moving code harder. It makes jumping around code blocks harder. Often the ide can help but sometimes it can’t.
In any curly-brace language these are things I simply don’t need to even think about. But in Python it’s a pain.
Yes it’s not the end of the world. Yes I can spend hours fine-tuning my editor. But… Why should I even have to? Why create these hurdles for no gain?
How do you “undo” whatever that script did?
And I still don’t care. Bad is bad even if a community is doing it.
Edit: Sorry if that was aggressive. This is a horrible practice and that community is the worst. They use HTTP by default? Encourage running scripts pointing to GH repositories controlled by community members? It’s just aching for the sort of supply-chain attacks we’re seeing with things like NPM has been enduring.