SansGuidon

https://morgan.zoemp.be my personal website, sort of a mixed blog and list

Helper scripts automate day 0. Cloudron automates day 2+.

Install ≠ operate

Proxmox = infra. You run ops. Cloudron = platform. Ops is mostly done. Clicking apps is easy. Maintaining them isn’t.

I still need to go back working on https://www.cinekids.info/ , a tool I made for myself to use before showing any movie to my 4 years old kiddo. It scrapes reviews from some parent friendly movie reviews websites and aggregates them.

I'm also automating more stuff around bookmarks management -> I used to manage an awesome list as a repository on GitHub for myself and over a couple of years there are relatively many stars on this repository. However I lost interest in maintaining this repo manually as I prefer to save my bookmarks on Shaarli. I'm coding a CLI tool to automate the work of syncing my shaarli links to my public "popular" (+500 stars) repo at https://github.com/SansGuidon/bookmarks

Myself and other users complain a lot about the "native" Plex -> Ombi watchlist integration being broken, I coded some sync tool to workaround the app malfunctions, by using Ombi, Plex and TMDB (The Movie Database) APIs and ensure Ombi is always up-to-date based on Plex watchlist. This works very well and allowed me to put a stop to the complains from my family members :-D

I'm also automating most of my email/linkedin interactions thanks to userscripts. And I keep automating more of the work I do around Cloudron, which is a very fun and stable platform to manage apps on VPS without the pains.

It becomes quite difficult nowadays to bookmark webpages or archive them, with CF, BotStopper/Anubis, Go Away etc... we will just need to all burn more cpu power just to access good content.

As a former user of N8N the tool looked interesting to me but I ended up converting most of my use cases into shell scripts, python scripts executed by cron jobs, and into ci/cd jobs. It gave me more flexibility about the tech stack I need, and a greater ease of debugging and developing robust designed tools.

I guess N8N was not intuitive for simple things and seemed too complicated for me. I'm now happier with cron jobs/GitOps to manage my automations. On the other hand I also had to replace some IFTTT workflows with my own scripts.

More work for me but I gained quality and control.

I mostly focus on text based content so PDF and webpages are easily supported. for PDFs I thought about using https://github.com/phiresky/ripgrep-all or pdfgrep https://pdfgrep.org/

For images, what do you want to grep for? for exif data -> https://exiftool.org/ if you want to find image based content, you might need something smarter. I think maybe it is a place where tools such as https://github.com/ultralytics/yolov5 can shine for me. simple enough to work with most of my images and tag them according to some preferences, and I would save such tags in a txt file.

Anyway, all metadata I store about images, links etc are all persisted in txt files. summaries, tags, etc, incoming/outgoing links etc, each has its own file. There are folders per link/content. Under each folder, one file per type of metadata. So it is very easy to know if some metadata is missing for a file, no index needed, it is just as simple as checking the presence of a file. everything is compatible with grep then.

for docx and xlsx it is out of my plate at this time, I didn't experiment enough to judge what works well enough. I hate those things.

Exactly... so much hype around complexity when simplicity wins. That's also why such systems like Wallabag, Linkwarden, Omnivore etc all disappointed me. In the end with a simple system made of static files and tools available out of the box on most distributions, I could make my own alternative to most archiving/bookmarking management systems and it just works. No DB, no framework, no fancy UI. Yet powerful. I have to blog about it.

Could be related to this as well -> https://github.com/linkwarden/linkwarden/issues/586 where content seems imported but its indexation is stuck in the queue. Blocker for me.

Even if importing them they might remain stuck in some import queue and you might not be able to search them. That was a blocker for me https://github.com/linkwarden/linkwarden/issues/586

I'm using Cloudron for 1.5 years and currently trying Yunohost and Coolify to try alternatives. But I gave up on Yunohost which felt too buggy and required too much workarounds and troubleshooting. I feel Coolify might be best for my needs but it's really for devs. Cloudron spoiled me so much everything else looks half baked.

Same here, I guess because I'm such a nerd, I have more affinities with like minded people that I often met at studies and at work. Other connections are fine but I'm not sharing the same dark humor and geek vibes with them. With WFH I still go to the office everyday and the people who I only see once a week seem too distant to me and we dont know each other very well.

It's now inactive-user on lobsters but it does not matter much anymore. I had changed my username few times on Lobsters.

Despite being a member of Lobsters for 5 years I've immediately disabled my account as soon as I felt the community becoming so hostile to brave users. I mean, I have no time to invest with a community that makes it so difficult to reach them and discuss any issue. Web is already painful enough.

the linked discussion shows more and more comments posted by inactive-users so users that have disabled their profile in reaction to lobste.rs decision and decided to leave lobste.rs for more open communities. I wish the best to lobste.rs but clearly they do not fight the right battle here and just close doors. Too bad for them.

this reminds me of a similar blog https://bradfrost.com/blog/post/just/

Typing fast is good but not so important as typing the right thing. Less is more and that's one of the key messages. There are degrees between typing fast and slow. I have the impression people are only able to see either extreme. The more we urge ourselves to eat fast, walk fast, do everything fast, the more anxiety we throw ourselves into, the more nervous we become when others are not fast enough, the less patience we develop, the more we tempt and bias ourselves to think more and fast is needed. There is no problem writing 1000 words per minute if we can so why would we think our text is repetitive ? After all we are good at typing, let's type while everyone is struggling to keep up with our rythme. Let's create more code to maintain, more code to debug, let's make big fonctions, long pull requests, let's be tempted to spam the chat and answer everything we can because we type fast enough. Let's skip the lunch because we will quickly answer them. Let's engage in more social networks because we have the bandwidth with our typing speed.

That's also a fact that the easiest an activity becomes the more we can become addicted to do it without thinking. The less we look for alternatives.

Why doing a meeting of 30min when 3000 lines of code can do it. Who cares who will maintain our code as long as we become the hero of the team by typing 10.000 loc overnight.

We are the faster. We win every discussion by typing more and faster. We impress. We throw more jira issues in backlog than everyone else. We dont see the need to slow down or refuse a task because our typing speed gives us more privilege and power.

So one key thing is becoming constrained also makes us more creative at problem solving. Maybe working more is not the key, maybe coding more is not the key goal. Maybe one line of code or a product can solve our problem. Maybe some llm, maybe our of our 100 ideas per minute, only one is relevant and maybe it won't be useful by next week. Good ideas are not always the newest ones.

I also consider on average programmers read more code than they write. I've often helped very busy people whose time was focused on reinventing the wheel and frameworks because they were proud of their coding skills and avoided looking elsewhere for existing solutions. I also haven't been the fastest programmer but always the one able to read code and find bugs better than peers. I'm not proud of typing more unnecessary and buggy code which will then slow us down through debugging and maintenance.

Observability, and quality is important, more than quantity and speed. And as for most things, speed also amplify some negativity, chaos and noise, in my experiences.

I'm not inclined to encourage multitasking just by rushing throwing more code while also driving while also writing a message to the kids while also eating. It's also charitable to take the time to focus. And maybe think about how this code is not worth it or not urgent anyway. And doing less can be good

I do agree on this (original blog author here)

For removing code or reading code, no need to be a fast typer and those activities are key part of learning and being productive

Thanks I find it not so hard already good results with only little code and time (thanks to llm for the draft).

I like the idea of OBTF and org mode but I find the syntax quite inconvenient. I do not like Markdown also. I'm currently taking inspiration from bujo, obtf, org mode and markdown for making my own thing based on characters I'm ok to type both on mobile and desktop and which can make it easy to find information back in a huge file both for humans and programs.

Typing is already easy enough. We overload the world with more texts and more code. We use AI-based summarizers anyway if too long, which is bad. Same for long code reviews, details are then overlooked. We write hundred pages essays to avoid the effort of slowing down and clarify our thinking. We neglect thinking and right to the point sentences because it's too easy to type fast and long, we feel the urge to react and to feel productive even if it's not. The urge to drive fast, to push more code, to write more. And put the pressure on everyone else to follow along with the melody. I would love more thinking and focused people, more careful too, and more aware people instead of having all of them wanting to type more and faster.

Link is back

It's back online. My vps provider is cheap for reasons...

The Belgian webmaster is aiming at moving away from the current tech stack and vps provider. For avoiding this kind of situation.

Original blog author here. The link should load now.

Git feels like solving puzzles everytime there are branches or collaboration made on a repo. So every day. Most productive days are when I dont need to use git. JJ is a workaround for git pains but the UX is not mature enough for replacing git yet.

Having distinct local working copies of same repo per git branch still works great, not stash needed, no conflicts or pain with local wip and drafts changes overwritten by pull. No blocker for switching branches. No reset hard/soft whatever. And makes files comparisons easier and makes my git workflows easier.

I still feel vcs could be simpler and solved problem if more engineers care about the topic of vcs and generally speaking about the ux of their main tools instead of creating new TUI and plugins to workaround broken ones.

The point for me is to build better and remove the need to know so much about vcs. Git is powerful but requires too much knowledge about git itself to get things done or to tackle possible conflicts in collaboration. Jujutsu tried to remove some of the concepts or make a better ui, to remove some complications. It's like providing an abstraction like python instead of C. Yet it's not perfect. I also struggle to transition to jujutsu yet I admit some benefits. But honestly both are yet too complicated and error prone for something as basic as communicate changes and align versions with each other. I'm happy to see initiatives such as jj but it's still an alpha.

Funny thing, I'm becoming an evangelist of Jujutsu but somehow I found myself stuck for 30 minutes with Jujutsu not being able to push a desired branch/commit, and Jujutsu had made my local git copy in detached HEAD. In the end I created a few useless branches through jj, and was still not able to articulate or understand through the docs how to sort my situation and tell jj that I want to move my latest work on main, nor to find a way to rebase/merge my accidentally created bookmarks/branches on top of main. Then ended up switching back to git, redo all the coding again and push. Having to understand either jj or git models through tons of tutorials and videos, while I just want a way to dialog easily with my changes in my own vocabulary and my own mental model, make me believe I'm a lost cause or I need my own VCS

Anyway Jujutsu stil feels great, I just hope now to not need to suffer like I did with git.