Hacker News Reader

Comment by sigjuice | Hacker News Reader

Docker is a bit of an overkill for such a tiny app IMHO

Oh yea it entirely is but I wanted to experiment with docker as ive never used it before, you can just pull the python and use export to set the vars if you want.

Edit: Docker is also an easy way to run in the background.

A bit of an overkill?

According to the developer, this is a Python script. So, rather than running on any desktop OS with minimal resources, it requires Docker, which requires a Linux Kernel. So for the vast majority of desktop users, this will require a Virtual Machine.

To send a fucking tweet.

This is the perfect example of developers doing dumb things, "because devops".

I've added instructions for pure python if you want it. I used docker because I've never used docker.

I've never used a concrete trowel to make waffle batter, doesn't mean I'm going to try it.

A bit of an overreaction don't you think? Especially after the OP explained that he is using docker for educational purposes and even included instructions for pure python. Seems unnecessary to continue ragging on him.

Using Docker to run a tool intended for home-use (i.e. not on a server) for 'educational purposes' is like learning to shave your bikini line by cutting the neighbourhood hedges.

Dude, I'm in highschool and wanted some experience with shit used in industry, gimme a break.

Not only 'learning Docker' is a pretty solid reason, it's not even that bad an idea. I have spun up containers for simpler stuff, like when I couldn't install a certain library because Anaconda messed up the install of a dependency which depended on native code.

I would say the risk reward on these two are a little bit different. With using Docker on something you are just trying out you risk wasting some time and maybe some abuse from stephenr on hackernews.

Practicing shaving your bikini line with hedgeclippers you stand to lose... well a lot more.

Notice I didn't say hedge cutters to shave.

I said cutting hedges. Vaguely related tasks (cutting something shorter) but vastly different tools, materials, scale, etc

This comment made me laugh so hard I got some weird looks from the person in the cube next to me.

On a more serious note, if you had never used a concrete trowel before and you weren't totally sure what it was, using it to make waffle batter would give you a better understanding of what it was and what it is good for.

I think packaging it up as a ready-to-use, cross-platform installable (which just happens to be in the docker container format) is better than expecting every desktop user to "install python3 python3-pip -y pip3 install pyspeedtest hurry.filesize tweepy". If the typical "desktop user" can't install one software with easily accessible installation instructions, is it reasonable to expect them to install 5 without?

I didn't say packaging is bad.

I said expecting desktop users to install docker to run a 35 line python script is fucking crazy.

You think someone who can't install python and 3 pip modules, is going to manage to install a Hypervisor, a Linux Guest, and Docker?

Maybe. We taught our support people how to install Docker via Docker Machine, and while that is resource-intensive and sometimes they come with questions, it did reduce the complexity of deploying new internal tools to them.

Perfect example where Go would be a great fit.

It uses Python 3 and several third party PIP installs, so nice for Mac users and people who are careful about running random unsandboxed code off the net.

It runs things as root inside Docker though, so not that much isolation from untrusted code.

Reply on news.ycombinator.com