Scott Hanselman

Lynx is dead - Long live Browsh for text-based internet browsing

July 13, '18 Comments [6] Posted in Docker | Open Source
Sponsored By

The standard for browsing the web over a text-=based terminal is Lynx, right? It's the legendary text web browser that you can read about at https://lynx.invisible-island.net/ or, even better, run right now with

docker run --rm -it nbrown/lynx lynx http://hanselman.com/

Awesome, right? But it's text. Lynx runs alt-text rather than images, and doesn't really take advantage of modern browser capabilities OR modern terminal capabilities.

Enter Browsh! https://www.brow.sh/

Browsh is a fully-modern text-based browser. It renders anything that a modern browser can; HTML5, CSS3, JS, video and even WebGL. Its main purpose is to be run on a remote server and accessed via SSH/Mosh

Imagine running your browser on a remote machine connected to full power while ssh'ing into your hosted browsh instance. I don't know about you, but my laptop is currently using 2 gigs of RAM for Chrome and it's basically just all fans. I might be able to get 12 hours of battery life if I hung out in tmux and used browsh! Not to mention the bandwidth savings. If I'm tethered or overseas on a 3G network, I can still get a great browsing experience and just barely sip data.

Browsing my blog with Browsh

You can even open new tabs! Check out the keybindings! You gotta try it. Works great on Windows 10 with the new console. Just run this one Docker command:

docker run -it --rm browsh/browsh

If you think this idea is silly, that's OK. I think it's brilliant and creative and exactly the kind of clever idea the internet needs. This solves an interesting browser in an interesting way...in fact it returns us back to the "dumb terminal" days, doesn't it?

There was a time when I my low-power machine waited for text from a refrigerator-sized machine. The fridge did the work and my terminal did the least.

Today my high-powered machine waits for text from another high-powered machine and then struggles to composite it all as 7 megs of JavaScript downloads from TheVerge.com. But I'm not bitter. ;)

Check out my podcast site on Browsh. Love it.

Tiny pixelated heads made with ASCII

If you agree that Browsh is amazing and special, consider donating! It's currently maintained by just one person and they just want $1000 a month on their Patreon to work on Browsh all the time! Go tell Tom on Twitter that you think is special, then give him some coins. What an exciting and artful project! I hope it continues!


Sponsor: Scale your Python for big data & big science with IntelĀ® Distribution for Python. Near-native code speed. Use with NumPy, SciPy & scikit-learn. Get it Today!

About Scott

Scott Hanselman is a former professor, former Chief Architect in finance, now speaker, consultant, father, diabetic, and Microsoft employee. He is a failed stand-up comic, a cornrower, and a book author.

facebook twitter subscribe
About   Newsletter
Sponsored By
Hosting By
Dedicated Windows Server Hosting by SherWeb
Monday, 16 July 2018 12:12:53 UTC
I'm sold. A few times I've been away from a power point and my Macbook lasts about half a day, the biggest drain is web browsing but it's a serious impediment to productivity if you don't have access to Google (and a few things besides). I found just switching off Chrome and using Safari sporadically gave me an extra couple of hours so brow.sh looks like it could be a big win.
Tuesday, 17 July 2018 05:41:47 UTC
I had thought that Links was the text standard, but browsh looks nice. Thanks for this. Now if selenium played nicely with these browsers.
Mark Nyon
Tuesday, 17 July 2018 07:46:35 UTC
How does it render the image as text without downloading the image? If it's downloading the image and processing it as text then does that really save that much power vs. just showing the image?
Peter Row
Tuesday, 17 July 2018 08:22:34 UTC
Peter Row above has a great point, and I have the same question regarding data saving. Doesn't it just download the full data of the text?

I mean I definitely see the use of this, but how it saves data, I can't tell.
Tuesday, 17 July 2018 09:13:19 UTC
To be fair, browsh stands on giants shoulders: all the dirty work is done by a Firefox run in headless mode, in a separated profile with custom webextension installed. Then browsh does the clever job of translating the Firefox-rendered page (to the detail of a single pixel) onto the TTY.

So this is indeed a great tool, but requires a lot of RAM and is relatively slow (esp. on startup).

https://www.brow.sh/docs/introduction/#design
Przemek
Saturday, 21 July 2018 05:46:51 UTC
Browsh is not technically a web browser. I went to their website. From the front page it says: "The only dependency is a recent 57+ version of Firefox."

And in the docs it says: "When the CLI starts, it looks for a compatible browser (currently only Firefox) and starts it in headless mode."

Browsh is just a wrapper between FF and the terminal. Kind of like a terminal front end for firefox, but you cannot run it on a true headless server since Firefox requires a graphics server e.g. Xorg Xephr Wayland Aqua or whatever Windows uses. Therefore it is not actually a browser. It cannot browse the web and interpret HTML, Javascript etc.

That doesn't mean all is lost. All those expectations of Browsh, would apply perfectly to Netsurf.

Netsurf can run in a terminal or a TTY via frame buffer (independent of any GUI toolkit or server) and it's based on it's own code, renders javascript etc. Their build system also sucks though and the only GUI toolkit they support for Linux is GTK as if there's not enough GTK/QT bloat pullers. However, build it with GTK disabled and frame buffer enabled, you'll have exactly what you wanted Browsh to be. Don't know anything about Windows framebuffer or if it has one. Servers generally do not run Windows on purpose anyways (does anyone?).

There's other lightweight browsers too; Netrider and Fifth use FLTK which is much more suitable to servers than GTK or QT. The only downfall is they use webkitGTK (ported to FLTK) which still pulls in a lot of typical GTK dependency bloat. Then there's Dillo which has potential if the developers did not convince themselves they are trying to suck, because it could work easily as a drop in replacement for the big three if they just fixed what remains for their CSS rendering and added a JS engine.
Trent
Comments are closed.

Disclaimer: The opinions expressed herein are my own personal opinions and do not represent my employer's view in any way.