11 KiB
How to choose your tools
This week I didn't take a look at HN to grab some news. And this week-end, in the morning I read those:
Similar articles have existed for years on different products. What is their common point? Software tooling and their potential change and disparition.
Accross the years, too many times I saw tools disapear. By tools I mean applications, web applications, web sites. I think we can also include programming languages, control versionning tools, building tools, package manager, etc…
The story can be quite different. Sometimes the disparition of a tool is positive, because I found a better one (from cvs to svn to git). But, too often, the tool simply disapears or worse downgrade its quality. I think we can find different names for those softwares:
- bloatware: remember digg, stumbleupon, windows?
- downgradeware: Swagger-UI v3 (v2 is neat), reddit new redesign (looks better, but slow)
- payware: Useful free service ask for money now. Or cost a lot more.
- crapware: Stop to works, quality degrate unless you pay: Twitter streaming API?
- dieware: Remember Friendfeed, Google Reader™, etc…
- etc…
This is often quite frustrating because you lose a lot of your investment with that tool.
Regarding Github Codespace; the integration of VSCode™ inside GitHub™ can be even worse. This is what I would call a trapware.
trapware: A software that is intented to put you inside a closed ecosystem. By slowly but surely add features that while looking great for the user at first sight will ensure to entrave other tools to interoperate.
Furthermore, the fact that Microsoft is involved give this story a taste of Embrace, Extend and Extinguish.
My real concern is that it could become a work framework. This could impose the full tooling on a lot of developers without giving them the freedom of choice.
For a startup CTO/CEO this GitHub™ Codespace™ could offer the following advantages:
- security: impossible or very hard to steal the source code by a single dev.
- homogeneity: all dev must use the same development environment. Thus the integration of new dev is faster.
- cheaper: don't need to pay for a full featured, fast machine to each new developer. A less performant machine able to display an electron app will do the trick.
- stats: you can observe the throughput of your developers. How many commits a day, how many lines of code, etc… How much bugs involved which part of the code and thus which dev to blame? How much time the dev is typing, moving its mouse, how much copy/paste is involved, etc…
For the single developers and open source developers this offer:
- homogeneity: if I learn how to work in this environment, I'll be easier to recruit and I'll know how to work fast.
- lower barrier to entry: for an opensource repository, it will become much easier for anyone to propose a PR to fix some issue. No need to local clone the project, no need to download all the dependencies to test it locally, etc…
But the price to pay is hidden.
- First, you are now, not able to choose your local working environment on your machine.
- GitHub™ can still change so much to become one of the previously
mentionned
/.*ware/you don't want to be involved with. They could forces you to pay a lot more, remove features, redesign to a bloatware, make it harder to interop with other platforms (prefer Azure to AWS etc…). - If everything involve machines in the cloud via the browser and via authorized plugins only. A lot of tools, features will never be allowed in this new ecosystem.
- Surveillance on meaningless or wrong metrics about your work. Instead of being evaluated on the feature you shipped or on other higher level metrics. It will be very tempting for your bosses to find flaws in your working habits. We are already living in a world were surveillance, metrics and stats are too easy to grab about a person. And anyone involved know this is all bullshit. Human are very good to play those kind of games. So people really working hard for the best will certainly perform badly compared to other people that simply trick the system.
So if the endgoal of GitHub™ is really to help open source and single developer. And more generally provide simply a better working experience by adding a new tool without any hidden marketing plan. Yes great. But I really doubt a company like Microsoft™ offer anything without a plan to make it worth it.
The Zoom acquires keybase is just another story of a dying product. Apparently the keybase team will probably stop maintaining keybase. The idea behind keybase was pretty nice. And they filled a gap in the current open source world.
The last article I mentionned was Making Emacs popular again. The first comment in HN was about how VSCode is easy to start with as compared to Emacs that need a lot more time to configure correctly for your needs. Yes, VSCode certainly just work and is easy to use. But Emacs is another beast. VSCode can become bad very fast, you don't control how it will evolve. The fact that this is a succesful Microsoft product does not garanty it will keep its currently quality. Emacs on the other hand is 44 year old and was designed so that it adapts to you. You are the one using libs and customizing it.
The argument to chose VSCode instead of Emacs look similar to me to the debate "Frameworks vs Libraries". Frameworks are easier to start with, but soon you find corner cases were you start to fight against them.
A Library on the other hand, is just a bunch of helpers you can use. And if you need another functionality, just make it using the libraries. But you have a lot more work to do yourself.
The common pattern I see during choice decision is often reducible to:
- Easy now, but less extensible and harder in the long run.
- Harder now, but more extensible and easier in the long run.
As a conclustion I would state that when you need to choose between different tools. Take the time to think about the investment costs. Sometime, the bit of pain in the begining is worth it. In particular if you are going to use this tool every days for many hours during the following years. If on the other hand you don't plan to use that tool much. Going with the easy option is certainly the best choice.
I consider Emacs to be of the 2nd option when compared to VSCode. Harder to start, but with a lot more control and potential power that you will probably never be able to get with most modern IDE/Editor. Also choosing a Free Software1 gives you a lot more control about its future.
Post-conclusion – Emacs is awesome
To go beyong my opinion, I'd like to share my experience with editors and emacs.
When I started to code.
We coded with vi, not vim, vi.
At that time I only knew, i, a, dd and cw vi commands.
So when I started to use IDEs I was thrilled.
After a few year I started to work for a company that forced me to use
their shitty computers.
I started to have wrist issues.
So I decided to learn vim.
And I saw the benefits only after a few weeks.
They were tremendous.
No more wrist pain.
And I started to learn a lot of editing automation.
Then I started a new work where we decided to code in Clojure. And so knowning that Clojure is a LISP and most LISPers love emacs because emacs plugin language is emacs LISP. I tried to use spacemacs. At that time I didn't want to invest much time in learning Emacs. I just wanted to learn the tricks that will make Emacs more valuable to my work. And it did after just a few days, weeks. I used Emacs superficially for years. Just Spacemacs + a few useful layers. And it was already quite efficient, at least as much as vim.
More recently I started to dig deeper. In particular, I read so much praise about org-mode I was really curious. And it took me some time to really discover why it is so great. First, let's just say that, basic org-mode is already quite valuable.
But you can do a lot. And unfortunately this is a bit hard to describe how org-mode is great without really digging a bit.
So you can think of org-mode as an extremely versatile todo-list / note taker with agenda and time tracking integration. Mostly you are in control of your working workflow with org-mode. The ability to do org-capture and org-refile is also great. Recently there is org-roam that is a step further to make orgmode a nice place to keep track of all your knowledge in one place.
Concretely, emacs has changed my workflow a lot and made me a lot more productive. It improved not only my coding workflow, but my full work environment. I started with the editor, a few plugins, and slowly, I integrated more aspect of my day to day tasks in emacs. Emacs is designed to adapt to your own needs you can start to automate a lot of small tasks.
I really love Emacs and if you want to joyfully join the Emacs users here are my advices:
Start by using either spacemacs or doom-emacs. It will take a few weeks to absorb vim keybindings. Slowly you'll start to learn how to configure it for your needs.
I really advise you to take a look at org-mode. Mastering it could change your carrier. Im my opinion org-mode alone is a good reason enough to use emacs. But there are a lot more to discover.
However, if you are used to tools from startups, with nice UI/UX. Almost no configuration cost. Be aware that digging in Free Softwares is a lot diffierent. Instead of having a big bundle with everything prepared to work you you will need to take the time to configure each part of a big system separately.
Howevery I'm deeply convinced the investment is really worth it.
note I said free software and not open source; c.f Why Open Source misses the point of Free Software