A simple GNU info replacement which isn't terrible.
This program displays the info pages as a single document in the standard pager
(usually less or more) with various redundant formatting and free software
propaganda removed (some pages contain more license text than actual useful
content).
You can install it with go get arp242.net/info, which will put the binary at
~/go/bin/info.
If the output is a terminal device the output will be piped to MANPAGER or
PAGER if set. The default is to use more -s.
INFOPATH is respected.
There are no commandline options at this point; all arguments starting with -
are ignored.
Is this too complex? You can approximate it with something like:
info() {
zcat "/usr/share/info/$1.info.gz" |
sed -Ee "/\x1f/d; /^File: $1.info,/d; /^(Node|Ref): .*�/d" |
cat -s |
less
}
This was the original version; but it was too hard to handle stuff like info pages split over several files (which this snippet won't handle) so I ended up writing the Go tool.
There's also info2man. I needed to make some changes to get it to work, and the output didn't look too great. This tool goes back to at least 2004. These criticism of Texinfo are hardly new, and the GNU folk's imperviousness to it isn't, either.
A Texinfo rant
I have been using several different variants of Unix for almost 20 years. Yet I still cannot navigate GNU's Texinfo.
-
Maybe the key bindings make sense from an Emacs perspective, but most of us aren't Emacs users. There is no way to make it behave like the more standard "vi-ish" key bindings as far as I can find. I need to learn an entire new set of key bindings. From the perspective of the majority of users the key bindings are just nonsensical.
If you're thinking "just read the manual you doofus": buzz off. Do you think I have nothing better to do than learn how to use some obscure piece of software I don't even like just to read some fecking plain text once a month? I've got about 80 years in my life, and this is not how I want to spend it.
And should the entire industry deal with this crap? How many man-hours will be wasted? Far too many. The only reason I sat down and wrote this in a bout of anger-driven development is the hope that it'll save myself and a bunch of other people some time and frustration.
-
I don't like tech docs split over loads of very small pages; only makes it harder to read or search for things.
I discovered that you can use
info --subnodes page | lessafter I wrote this tool to output to a single page. I originally tried this, but it doesn't work unless you pipe it; justinfo --subnodesbehaves as usual and the option is silently ignored 🤦It won't strip all the crap though (like now-useless navigation).
-
There is no longer a single source for documentation; do I look at the info page or man page? Some tools have both, but one is more comprehensive than the other (sometimes info, sometimes man).
Fragmentation and inconsistency has long been considered one of Unix's weak points, and rightfully so. Texinfo only makes this worse. Then again, "GNU's not Unix" 🤷
-
Every page comes with an inordinate amount of free software advocacy. I want to read a bloody manual, not a political statement. I know where the websites of the FSF and GNU are if I'm interested.
Example: tar page minus crap is reduced from 91k words to 73k, grep from 7.5k to 14.5k. That's right, over half of the grep info page is Free Software wankery ✊🍆 💦
-
Some pages include images, which don't work inside a terminal (all of the images in the info pages on my system could be easily represented with a simple ASCII diagram, so why use images? Because you can I guess.)
There are exactly two good points:
-
It's not widely used outside of GNU.
-
The source format is fairly readable, so it's not too hard to write an alternative reader for it.
The entire Texinfo project is a failure. A more reasonable person would have moved on from it a decade ago.
None of this means that man pages couldn't do with some enhancement – better referencing probably being the most prominent – but GNU info isn't the answer.
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent