Workflow
2014-05-31
The inspiration for this came from this blog post, where the author describes how he uses his computer. While he does use CRUX, a GNU/Linux distro and I use OpenBSD, our workflows are actually surprisingly similar, which can, in part, be attributed to the
inspiration for the ports system both CRUX and OpenBSD took from FreeBSD, although OpenBSD is obviously a lot closer to the original.
Using OpenBSD
OpenBSD is a very traditional Unix-like OS that doesn't get in your way or do anything for you. In contrast to the author of the blog post I linked to, I appreciate a good set of defaults and a smooth install. Indeed, you can install OpenBSD by mashing enter and only stopping to enter a root password, and you'll end up with a very sane set of defaults - an X server, some X programs, tmux, and not to mention the other software that comes in the base system - an FTP server, Nginx, and so on.
Even with all of this "bloat", the install footprint is well under a gigabyte and RAM usage (checked from the text console) stays under 32 MB. So either most GNU/Linux distros are using all that disk space and RAM to do things I don't want or need, or they do the minimum, but a lot less efficiently than OpenBSD.
To summarise, if you want to use OpenBSD, here are the steps to follow:
- Download an install image from here
- Boot it from USB, CD or wherever
- Hit enter until it tells you to stop
- Reboot
- Install packages
After that, it's really identical to using Solaris, AIX or another BSD - OpenBSD is fully POSIX compliant. That does, however, mean that it's slightly different from the GNU userland (cp -R instead of cp -r, for instance).
Note that I didn't mention ports: the idea with OpenBSD is that you use the packages - the maintainers put a lot of effort into making sure you don't need to compile packages from ports yourself. Largely, every flavour of package you want already exists and has been compiled. Maybe you want vim with Perl, Python and Ruby support - good news! There's a package with that compiled in. You want rsync to use libiconv? They've accounted for that use case too! Ports should only be downloaded and used in very niche cases.
Window Manager
I'm not very adventurous with my WM, I just use i3, a very popular WM. I say that like it's a bad thing, but there's a reason it's popular - it does what it says on the tin very well. The default keybindings are fine for me, although I use Mod4 (super) instead of Mod1 (alt). You can find more about it here, and to install it, you should be able to find it in your OS's repos. It's in OpenBSD under "i3", unsurprisingly.
For the uninitiated, it's a tiling window manager, which means that, most of the time, my windows take up the whole screen and when there are multiple, they "tile", meaning they are sized such that there are no gaps and no windows behind another. Here's a GIF of i3 in action:
Web Browsing
This is the activity I spend most of my time doing. Nowadays, I use luakit, which is a Webkit-based browser with Vim-like keybindings, mostly based on Pentadactyl, a popular Firefox plugin I used to use.
Checking Email
The activity I spend the second-most amount of time on while on my laptop is checking my mail. I'm subscribed to a few mailing lists, but the one I check the most is misc@openbsd.org. A close second is tech@openbsd.org. While these aren't overly high-volume lists, it takes a while for each message to be fetched from the server, and this must be done one-by-one as I view them if I were to use IMAP.
Those seconds add up. Maybe they don't, but waiting a second between
reading messages is very annoying, so my only option was to store the
messages locally, allowing me to read them with no network delay.
Setting this up was easy, I just used the same Mutt config as I used
when on IMAP and configured it for local mail in a Maildir located at
~/mail/. Even better, this meant I could configure my system MTA to
deliver all system mail there, too (cron job errors, security output,
nightly backups).
Here are the relevant lines from my .muttrc:
set mbox_type=Maildir
set spoolfile = "~/mail/"
set folder = "~/mail/"
To actually fetch the mail, I used getmail, a small Python script (can
be installed with pkg_add getmail). Setting that up was very simple,
too, just install and place the following in .getmail/getmailrc:
[retriever]
type = SimplePOP3SSLRetriever
server = pop.example.com
username = username
password = password
[destination]
type = Maildir
path = ~/mail/
user = yourusername
[options]
read_all = false
Getmail can be run by simply running getmail, but this is very
annoying to do every time you want to read mail, and sort of defeats the
point - this was supposed to save time. I set up a cron job for it -
place the following in the file opened via crontab -e:
30 * * * * getmail
That runs getmail every 30 minutes, meaning your mail stays up to date.
If you're ever waiting for an important message, you can still run
getmail to refresh your inbox directly, although it is mildly annoying
to open your shell every time you want to do that - why not bind it to a
key in Mutt?
To do that, add the following to your .muttrc:
macro index "^" "!getmail<enter>"
That means that, whenever in the index, you can press "^" to get any new mail.
No place like ~
You already know I have a ~/mail/ directory, but I have a few more:
~ >> tree -L 2
.
|-- bin -> etc/scripts/bin
|-- etc
| |-- README.md
| |-- [...]
| `-- zsh
|-- mail
| |-- cur
| |-- new
| `-- tmp
|-- notes
|-- sent
|-- src
| |-- advent
| |-- [...]
| `-- www
`-- var
|-- documents
`-- music
47 directories, 4 files
As you can see, I keep my source code in ~/src/, data files in
~/var/, mail in ~/mail/, scripts in ~/bin/ and configs in
~/etc/. My etc directory is filled with configs which are symlinked to
from their expected locations. I'm sure I've written a post about it at
some point.
Just like most people, I'm a fan of text files, but not any ordinary text files - most of my notes are written using Emacs' org-mode, which I believe is a note-taking environment second to none. Unless I'm programming in a REPL language (Ruby, Python, Lisp), I tend to use Vim. I favour Emacs for those languages because...well, you have to use it to really understand what I mean - look up some cool things you can do with SLIME.
You can check out my dotfiles here.