Power Overwhelming - Hacking

Hangul spellcheck for Vim

2024-10-31T13:37:00-04:00

There’s got to be a better way to do this. Someone please enlighten me.

Modern Korean is written in 한글 (Hangul), which uses a syllabic alphabet. It includes spaces between words, unlike Chinese or Japanese, which means that it’s possible to have meaningful spellchecking.

So of course one day I decided I wanted to configure Vim to support spellchecking Hangul. Unfortunately, there’s no file ko.utf-8.spl at ftp.vim.org, and in a cursory search I couldn’t find an.

On the other hand, the hunspell tool does have a Korean dictionary, and there’s a PKGFILE provided for ARCH, so by running pikaur -S hunspell-ko I was able to obtain the files

/usr/share/hunspell/ko_KR.aff
/usr/share/hunspell/ko_KR.dic.

In theory, if you then run the Vim command

:mkspell /tmp/ko /usr/share/hunspell/ko_KR

then Vim would create the file /tmp/ko.utf-8.spl, which you can then place into ~/.vim/spell and get spellchecking.

Unfortunately, theory is not the same as practice, by which I mean that the process threw a bunch of warnings and the resulting file it totally didn’t work — every word was being marked as misspelled.

So I spent a bit of time trying to debug it, and being like, “how come these two words that look exactly the same are showing up as different?”

It turns out there’s actually (at least) two different ways to encode of Hangul into Unicode, namely NFD (decomposed) and NFC (composed). The input program I’m using produces NFC glyphs, where each Hangul syllable block is a single code point; but the spellcheck file /usr/share/hunspell/ko_KR.dic instead has entries in NFD, where each atom within the syllable block is a character.

That diff is spooky.

Actually, NFD makes sense for a spellchecker, because you’d want something like 한 (Romanized han) and 항 (Romanized hang) to have an edit distance of 1/3 rather than 1. Then, in order to deal with NFC inputs, the /usr/share/hunspell/ko_KR.aff provides many ICONV and OCONV directives that tell the spellchecker how to convert from NFC input into NFD and then back. So hunspell works well.

The problem is that Vim’s :mkspell command apparently doesn’t actually support ICONV and OCONV. In order to force it to work, I ended up just writing a Python script that stripped all the unsupported commands from ko_KR.aff, and converted ko_KR.dic into NFC format.

import unicodedata

UNSUPPORTED_WORDS = (
    "LANG",
    "WORDCHARS",
    "ICONV",
    "OCONV",
    "AF",
    "MAXCPDSUGS",
    "MAXNGRAMSUGS",
    "MAXDIFF",
    "COMPOUNDMORESUFFIXES",
)

# Make the aff file but take out things unsupported by Vim
with open("/usr/share/hunspell/ko_KR.aff", "r", encoding="utf-8") as infile, open(
    "ko_KR.aff", "w", encoding="utf-8"
) as outfile:
    for line in infile:
        if not any(line.startswith(word) for word in UNSUPPORTED_WORDS):
            print(line.strip(), file=outfile)

# Make the dic file but re-normalize it to NFC
with open("/usr/share/hunspell/ko_KR.dic", "r", encoding="utf-8") as infile:
    content = infile.read()
content = unicodedata.normalize("NFC", content)
with open("ko_KR.dic", "w", encoding="utf-8") as outfile:
    print(content, file=outfile)

After storing these mutilated files into ~/dotfiles/vim/spell/korean-setup/ko_KR and running :mkspell /tmp/ko ~/dotfiles/vim/spell/korean-setup/ko_KR, the outputted /tmp/ko.utf-8.spl can now check for spelling errors (at least if the words are in NFC format).

It’s working. Sorta.

A summary of this process is posted on my dotfiles GitHub. To actually use this, just download the ko.utf-8.spl file directly, no need to re-follow the steps.

The issue is that because the spellcheck dictionary is using NFC now, while it can highlight the red words, the “suggestions” provided aren’t particularly good. If you look at the suggestions for the spellcheck for the typo’d word 항글, the top 10 are:

 1 "한글"
 2 "고글"
 3 "궁글"
 4 "담글"
 5 "답글"
 6 "댓글"
 7 "덧글"
 8 "동글"
 9 "둥글"
10 "빙글"

The problem is that because NFD encodes the characters by block, any change to the entire first syllable cause an equally bad edit distance. So while I’ve managed to get quick highlighting of mistakes, the autocorrection of those mistakes isn’t really there.

I feel like this whole process was a convoluted hack. Is there a better way to do this?

Why Evan does not like JavaScript as a first language

2022-11-18T13:37:00-05:00

Some people have asked me why I anti-recommend JavaScript for beginners on my website FAQ. This post will try to give a few reasons why.

Some notes:

I’m referring to base JS. I like TypeScript a lot for example (it’s high on my tier list of programming languages for beginners). And I know about eslint, but you asked me about recommendations for beginners, and I think beginners shouldn’t have to worry about setting up an IDE with strict linting until after they can write a for loop by themselves without screwing up.
I have multi-file projects in mind. I don’t have a problem with using inline JavaScript for tiny 20-line snippets of code embedded in a webpage.
I’m an amateur programmer. Professional programming is a different ball game.

Weak typing

I’m clumsy. I make many more mistakes than the average programmer. The other day I finally installed a Django-HTML linter in OTIS-WEB and found out that the majority of my templates had mismatched HTML tags I never noticed for years.

If I write some nonsense code like

let x = 3;
let y = x.value;

I would like the program to, you know, crash. It doesn’t. It sets y = undefined. And then 100 lines later I have to figure out why some other function didn’t do what it was supposed to, and trace back through the entire source.

It’s at a point where you can generate memes by just taking two different types of objects and adding or subtracting them and then laughing at the result.

Welcome to Node.js v19.1.0.
Type ".help" for more information.
> '7' + 3
'73'
> '7' - 3
4
> 'hi' + ['i', 'am', 'a', 'potato']
'hii,am,a,potato'
> const x = 42;
undefined
> '1337' + x - x
133700
> '1337' - x + x
1337
> 'wtf' - 2022 + 'i?'
'NaNi?'

Like, at least in Python, you’ll get a TypeError or AttributeError or something when you run the code (but not at “compile time”, because scripting languages don’t have a “compile time”). Python’s my native language, and it still annoyed me so much that I use both mypy and pyright on all my Python code now to enforce what essentially became Python with static types, despite Python’s type system being janky af. You better believe I’m doing this with JavaScript too.

null, undefined, NaN

These keywords strike dread into my heart because I lack the professional training to keep the differences straight in my head.

Last I checked though:

Welcome to Node.js v19.1.0.
Type ".help" for more information.
> null === null
true
> undefined === undefined
true
> NaN === NaN
false

Oh yeah, for those of you that don’t know, you always use === and not == these days because the latter operator is a total minefield:

> 0 == false
true
> 0 == null
false
> 0 == undefined
false
> 0 == !!null
true
> 0 == !!undefined
true
> [] == 0
true
> const a = new Number(42);
undefined
> const b = new Number(42);
undefined
> a == b
false
> a == 42 && b == 42
true

Well, == is supposed to be symmetric at least. Except apparently in Internet Explorer, where window == document is true but document == window is false, in some versions anyway.

Variable scope

It used to be that you could forget a var keyword and then suddenly your variable became a global variable. These days we have let and const and we just tell the beginners to always use one of these two and never use var (or worse no keyword at all) so I guess it’s not as bad as it used to be.

Though of course you can still find some contrived examples.

Welcome to Node.js v19.1.0.
Type ".help" for more information.
> (function() { const x = y = 1; })();
undefined
> x
Uncaught ReferenceError: x is not defined
> y
1

(For those of you that don’t know, the whole (function() {})() construction is an actual JavaScript idiom that’s used to prevent exactly this situation.)

And the this keyword still strikes fear into my heart. I can’t understand it for the life of me.

Modules and classes

The JavaScript base language used to have no module system. Now we have CommonJS and ES modules instead. Which I guess is fine, but it would have been nice to have these in the base language, you know? (I’m definitely not saying this because I spent an entire night fighting some node dependency chain where one component used CommonJS, and the other used ES, and they wouldn’t play along.)

These days I think people just use npm or something and be glad that with Python I have standard library that won’t collapse one day because left-pad disappeared suddenly.

Similarly, JavaScript uses a prototype-based object system that I never really understood either, because it looks like the classes I’m used to, but isn’t. If I was a professional programmer I’d probably take a few hours or days out of my life to figure out what the heck was going on. I’m not, and whatever beginner is reading my FAQ is definitely not.

Easter eggs

Actually these might not be a downside 🙂. I like Easter eggs. Maybe not in day-to-day code. But examples like this make for a good laugh, and they’re harmless because if you actually run into them in real life you’re doing something wrong anyway.

Welcome to Node.js v19.1.0.
Type ".help" for more information.
> null > 0
false
> null == 0
false
> null >= 0
true
> 100 <= 200 <= 300
true
> 300 >= 200 >= 100
false
> regex = /nani/g;
/nani/g
> regex.test('nani?');
true
> regex.test('nani?');
false
> [7, 12, 13].sort()
[ 12, 13, 7 ]

See wtfjs.com for more.

Summary

JavaScript is the language that runs the web, so it’s kinda mandatory. If you want client-side code to run everywhere, it needs to be in a browser, so it has to be JavaScript.

But I still think people should learn how to program first, and then learn JavaScript, rather than the other way around. Or, at least start with one of the gazillion languages that transpiles to JavaScript so that you don’t have to fight language quirks and linters until after your feet are at least moderately wet.

SysRq on Arch Linux Mac Mini

2016-12-14T13:37:00-05:00

This post documents my adventures of getting the SysRQ key working on my Mac Mini and Macbook (both running Arch Linux). The suggestions of loadkeys and keyfuzz that are the first search entries don’t work for me, so some more sophisticated black magic was necessary.

Remapping the Fn keys

This step is technically optional, but I did it because the function keys are a pain anyways. Normally on Apple keyboards one needs to use the Fn key to get the normal Fn keys to behave as a F<n> keystroke. I prefer to reverse this behavior, so that the SysRq combinations is Alt+F13+F rather than Fn+Alt+F13+F, say.

For this, the advice on the Arch Wiki worked, although it is not thorough on some points that I think should’ve been said. On newer kernels, one does this by creating the file /etc/modprobe.d/hid_apple.conf and writing

options hid_apple fnmode=2

Then I edited the file /etc/mkinitcpio.conf to include the new file:

...
BINARIES=""

# FILES
# This setting is similar to BINARIES above, however, files are added
# as-is and are not parsed in any way.  This is useful for config files.
FILES="/etc/modprobe.d/hid_apple.conf"

# HOOKS
...

Finally, recompile the kernel for this change to take effect. On Arch Linux one can just do this by issuing the command

$ sudo pacman -S linux

which will reinstall the entire kernel.

Obtaining the keystroke

Next, I needed to get the scancode of the key I wanted to turn into the SysRQ key. For me attempting showkey -s did not work so I instead had to use evtest, as described in this Arch Wiki.

$ sudo pacman -S evtest
$ sudo evtest
No device specified, trying to scan all of /dev/input/event*
Available devices:
/dev/input/event0:  Logitech USB Receiver
/dev/input/event1:  Logitech USB Receiver
/dev/input/event2:  Apple, Inc Apple Keyboard
/dev/input/event3:  Apple, Inc Apple Keyboard
/dev/input/event4:  Apple Computer, Inc. IR Receiver
/dev/input/event5:  HDA NVidia Headphone
/dev/input/event6:  HDA NVidia HDMI/DP,pcm=3
/dev/input/event7:  Power Button
/dev/input/event8:  Sleep Button
/dev/input/event9:  Power Button
/dev/input/event10: Video Bus
/dev/input/event11: PC Speaker
/dev/input/event12: HDA NVidia HDMI/DP,pcm=7
/dev/input/event13: HDA NVidia HDMI/DP,pcm=8
Select the device event number [0-13]: 2
Input driver version is 1.0.1
Input device ID: bus 0x3 vendor 0x5ac product 0x220 version 0x111
Input device name: "Apple, Inc Apple Keyboard"

This is on my Mac Mini; the list of devices looks different on my laptop. After this pressing the desired key yields something which looked like

Event: time 1456870457.844237, -------------- SYN_REPORT ------------
Event: time 1456870457.924097, type 4 (EV_MSC), code 4 (MSC_SCAN), value 70068
Event: time 1456870457.924097, type 1 (EV_KEY), code 183 (KEY_F13), value 1

This is the F13 key which I want to map into a SysRq — the keycode 70068 above (which is in fact a hex code) is the one I wanted.

Using udev

Now that I had the scancode, I cd’ed to /etc/udev/hwdb.d and added a file 90-keyboard-sysrq.hwdb with the content

evdev:input:b0003*
  KEYBOARD_KEY_70068=sysrq

One then updates hwdb.bin by running the command

$ sudo udevadm hwdb --update
$ sudo udevadm trigger

The latter command makes the changes take effect immediately. You should be able to test this by running sudo evtest again; evtest should now report the new keycode (but the same scancode).

One can test the SysRQ key by running Alt+SysRq+H, and then checking the dmesg output to see if anything happened:

$ dmesg | tail -n 1
[  283.001240] sysrq: SysRq : HELP : loglevel(0-9) reboot(b) crash(c) ...

Enable SysRq

It remains to actually enable SysRQ, according to the bitmask described here. My system default was apparently 16:

$ sysctl kernel.sysrq
kernel.sysrq = 16

For my purposes, I then edited /etc/sysctl.d/99-sysctl.conf and added the line

kernel.sysrq=254

This gave me everything except the nicing of real-time tasks. Of course the choice of value here is just personal preference.

Personally, my main use for this is killing Chromium, which has a bad habit of freezing up my computer (especially if Firefox is open too). I remedy the situation by repeatedly running Alt+SysRq+F to kill off the memory hogs. If this doesn’t work, just Alt+SysRq+K kills off all the processes in the current TTY.

DNSCrypt Setup with PDNSD

2016-11-10T13:37:00-05:00

Here are notes for setting up DNSCrypt on Arch Linux, using pdnsd as a DNS cache, assuming the use of NetworkManager. I needed it one day since the network I was using blocked traffic to external DNS servers (parental controls), and the DNS server provided had an outdated entry for hmmt.co. (My dad then pointed out to me I could have just hard-coded the necessary IP address in /etc/hosts, oops.)

For the whole process, useful commands to test with are:

nslookup hmmt.co will tell you the IP used and the server from which it came.
dig www.hmmt.co gives much more detailed information to this effect. (From bind-tools.)
dig @127.0.0.1 www.hmmt.co lets you query a specific DNS server (in this case 127.0.0.1).
drill @127.0.0.1 www.hmmt.co behaves similarly.

First, pacman -S pdnsd dnscrypt-proxy (with sudo ostensibly, but I’ll leave that out here and henceforth).

Run systemctl edit dnscrypt-proxy.socket and fill in override.conf with

[Socket]
ListenStream=
ListenDatagram=
ListenStream=127.0.0.1:40
ListenDatagram=127.0.0.1:40

Optionally, one can also specify which server which DNS serve to use with systemctl edit dnscrypt-proxy.service. For example for cs-uswest I write

[Service]
ExecStart=
ExecStart=/usr/bin/dnscrypt-proxy
     -R cs-uswest

The empty ExecStart= is necessary, since otherwise systemctl will complain about multiple ExecStart commands.

This thus configures dnscrypt-proxy to listen on 127.0.0.1, port 40.

Now we configure pdnsd to listen on port 53 (default) for cache, and relay cache misses to dnscrypt-proxy. This is accomplished by using the following for /etc/pdnsd.conf:

global {
    perm_cache = 1024;
    cache_dir = "/var/cache/pdnsd";
    run_as = "pdnsd";
    server_ip = 127.0.0.1;
    status_ctl = on;
    query_method = udp_tcp;
    min_ttl = 15m;       # Retain cached entries at least 15 minutes.
    max_ttl = 1w;        # One week.
    timeout = 10;        # Global timeout option (10 seconds).
    neg_domain_pol = on;
    udpbufsize = 1024;   # Upper limit on the size of UDP messages.
}

server {
    label = "dnscrypt-proxy";
    ip = 127.0.0.1;
    port = 40;
    timeout = 4;
    proxy_only = on;
}

source {
    owner = localhost;
    file = "/etc/hosts";
}

Now it remains to change the DNS server from whatever default is used into 127.0.0.1. For NetworkManager users, it is necessary to edit /etc/NetworkManager/NetworkManager.conf to prevent it from overriding this file:

[main]
...
dns=none

This will cause resolv.conf to be written as an empty file by NetworkManager: in this case, the default 127.0.0.1 is used as the nameserver, which is what we want.

Needless to say, one finishes with

systemctl enable dnscrypt-proxy
systemctl start dnscrypt-proxy
systemctl enable pdnsd
systemctl start pdnsd

Shifting PDF's using gs

2016-04-03T13:37:00-04:00

Some time ago I was reading the 18.785 analytic NT notes to try and figure out some sections of Davenport that I couldn’t understand. These notes looked nice enough that I decided I should probably print them out, But much to my annoyance, I found that almost all the top margins were too tiny, and the bottom margins too big. (I say “almost all” since the lectures 19 and 24 (Bombieri proof and elliptic curves) were totally fine, for inexplicable reasons).

Thus, instead of reading Davenport like I told myself to, I ended up learning enough GhostScript flags to write the following short script, which I’m going to share today so that other people can find better things to do with their time.

#!/bin/bash
for file in $@
  do
    echo "Shifting $file ..."
    gs \
      -sDEVICE=pdfwrite \
      -o shifted-$file \
 -dPDFSETTINGS=/prepress \
 -c "<</PageOffset [0 -36]>> setpagedevice" \
 -f $file
done

The arguments 0 and -36 indicate to not change the left/right margins, but to shift the content vertically downwards by 36pt (a half inch). Of course, they can and should be adjusted depending on specific task. Invocation is the standard ./script-name.sh *.pdf (or whatever).

(Aside: ironically, this decreased the file sizes of the affected PDF’s.)

Git Aliases

2015-10-25T13:37:00-04:00

For Git users:

I’ve recently discovered the joy that is git aliases, courtesy of this blog post. To return to the favor, I thought I’d share the ones that I came up with.

For those of you that don’t already know, Git allows you to make aliases – shortcuts for commands. Specifically, if you add the following lines to your .gitconfig:

[alias]
    cm = commit
    co = checkout
    br = branch

Then running git cm will expand as git commit, git co master is git checkout master, and so on. You can see how this might make you happy because it could save a few keystrokes. But I think it’s more useful than that – let me share what I did.

The first thing I did was add

pu = pull origin
psh = push origin

and permanently save myself the frustration of forgetting to type origin. Not bad. Even more helpful was the command

undo = reset --soft HEAD~1

Thus if I make a commit and then decide I want to undo it, rather than having to remember (or Google) what the correct incantations were, I just have to type git undo. It’s really an undo button!

Now for the fun part – some of Git’s useful commands are pretty verbose and take up lots of space. For example, here’s what git status looks like:

git status

Kind of verbose if you ask me, and by now I know what “git pull” does. Fortunately, it turns out that there are some options you can run to make this look nicer. All you have to do is say git status -s -b, or in the context of this post, set the alias

ss = status -s -b

Then you get

git ss

which is much cooler.

Similarly, git log takes up a lot of space. I have the following format, which I’ve edited from the above blog post to suit my own tastes.

ls = log -n 16 --pretty=format:"%C(yellow)%h\\ %C(cyan)[%cn]\\ %C(reset)%s\\ %C(red)%d" --decorate
ll = log -n 6 --pretty=format:"%C(yellow)%h\\ %C(cyan)[%cn]\\ %C(reset)%s\\ %C(red)%ad" --decorate --date=short --stat

These give in my opinion the much more readable format

git ls and git ll

If you’re on a branch that does merges, you might also have fun with

tree = log -n 16 --pretty=format:"%C(yellow)%h\\ %C(cyan)[%cn]\\ %C(reset)%s\\ %C(red)%d" --decorate --graph

which will put these into a graphical tree for your viewing pleasure.

And finally a few more that I find nice, some again taken directly from the link above:

fail = commit --amend # to avoid stupid "oops typo" commits
rb = rebase
rbc = rebase --continue
bis = bisect
dc = checkout -
assume = update-index --assume-unchanged
unassume = update-index --no-assume-unchanged
assumed = "!git ls-files -v | grep ^h | cut -c 3-"

(Here “dc” is short for “discard”, since git dc file discards the changes to that file.) And that’s just the beginning of what you can do!

Preemptive answer: I’m also using git-completion (for tab-completing in git) and git-prompt with the line

export PS1='[\033[0;32m]${debian_chroot:+($debian_chroot)}\u@\h [\033[0;33m]\w$(__git_ps1 " [\033[1;31m]#%s")\n[\033[0m]\$ '

in my bashrc. That’s where the branch indicators are coming from. The terminal is XFCE4.

Arch Linux on a Mac Mini

2014-10-07T13:37:00-04:00

This post briefly outlines the process of setting up a dual boot OSX and Arch Linux on a Mac Mini. This is mostly for my reference in the likely event that I will be doing anything similar in some years, so it assumes some competence; fortunately, the Arch Wiki’s Beginner’s Guide probably fills in any gaps I left out. Obligatory Disclaimer: Use at your own risk or not at all.

This is almost the same as any other installation of Arch Linux, with a few changes that took some hours of frustration to figure out because of the EFI booter. My method is to create the partitions in Disk Utility, install rEFInd, and then install the grub bootloader into /dev/sda1.

Setup done in OSX

First, install rEFInd. This worked out of the box for me, and makes it possible to boot via USB.
Create a Arch Linux installer USB by dd-ing (or anything else) the latest installation medium onto a USB drive.
Set up the partitions; I find it easier (and less dangerous) to just use the OSX Disk Utility to do this. See, for example, the procedure here. My OSX installations appear to come with three partitions, a small one called “EFI”, a main “OS X HD” partition, and then a small “Recovery HD”, like so:
```
  NAME LABEL TYPE SIZE
  sda disk 931.5G
  |-sda1 EFI part 200M
  |-sda2 OS X HD part 927.9G
  `-sda3 Recovery HD part 619.9M
```
(This output is from lsblk, and is not what Disk Utility looks like). I like to create a partition for my Arch Linux system (which I name “Arch”) and a fifth partition just for the /home directory (which I name Home”). This leaves me to something like
```
  NAME LABEL TYPE SIZE
  sda disk 931.5G
  |-sda1 EFI part 200M
  |-sda2 OS X HD part 476.9G
  |-sda3 Recovery HD part 619.9M
  |-sda4 Arch part 179G
  `-sda5 Home part 272G
```

Booting into the USB and finishing up the partitions

Now that the partitions and rEFInd is set up, and the USB is written, we can proceed with the actual installation. At this point, one can basically follow the standard procedure with a few changes.

Reboot the device into the USB. Since rEFInd is installed, it should give you the option of booting into the USB.
Establish an Internet as required.
We’ve already created the partitions in Disk Utility above, so there is no need to change the partitions themselves now. However, it is necessary to format the newcly created partitions above. In my case, the relevant commands are
```
  # mkfs.ext4 /dev/sda4
  # mkfs.ext4 /dev/sda5
```
Warning: Please, please make sure that you are formatting the right partitions. The command lsblk -f will print out the partitions and their labels.
Now we need to mount the directories. There are three directories we need to mount, the main filesystem and the home directory, as well as the EFI boot directory. The part that was non-obvious to me is that the boot directory we want is actually the “EFI” directory (likely /dev/sda1) that OSX already provides. The relevant commands in my case were
```
  # mount /dev/sda4 /mnt
  # mkdir /mnt/home
  # mount /dev/sda5 /mnt/home
  # mkdir /mnt/boot
  # mount /dev/sda1 /mnt/boot
```
Now you can happily install the base system and generate an fstab file:
```
  # genfstab -U -p /mnt >> /mnt/etc/fstab
```

Configuring the base system and installing the bootloader

Now we can chroot into the system and follow all the directions, up to (but not including) installing the bootloader.

I could not get gummiboot to work but maybe you will have better luck. Fortunately, with /dev/sda1/ mounted as /boot, I got GRUB to work nicely.

  # pacman -S grub
  # grub-install --target=x86_64-efi --efi-directory=/boot --bootloader-id=arch_grub --recheck
  # grub-mkconfig -o /boot/grub/grub.cfg

Now we can exit the chroot environment and power down the system.

If all goes well, upon rebooting, rEFInd will now boot into the complete Arch Linux system.

PDF Compression

2014-02-01T13:37:00-05:00

I always scan copies of letters into my computer before I send them out. So I had a bunch of large PDF’s sitting around hogging my Dropbox space.

One day I found this blog post which claimed that simply running (in Bash) the commands

$ pdf2ps original.pdf temp.ps
$ ps2pdf temp.ps new.pdf

would decrease the file size. (The two commands are part of GhostScript, which I had installed on my Linux boxes anyways.) I couldn’t resist trying it – and miraculously, it worked. It generally decreases my scans by a factor of 10 (from 20MB to 2MB or so).

I have no clue why this works, although it probably has something to do with the fact that the PDF’s are scanned pages . Anyone care to enlighten me?