diff --git a/homepage/blog b/homepage/blog
index 4fdf7ed..e0c0e00 100755
--- a/homepage/blog
+++ b/homepage/blog
@@ -37,6 +37,150 @@ __NAVIGATION__
 </PRE></BODY></HTML>
 
 
+2022-02-08
+
+	If you had ghosts in your blood cocaine would totally work on getting
+rid of the ghosts.
+
+
+2022-02-07
+
+#!/bin/sh
+set -ex;mkdir -p blah;python -c "import os;os.chdir('blah')
+with open('../$0', 'r') as f:
+	for day in f.read().split('\n\n\n'):
+		if day.split('\n')[0] == '#!/bin/sh':
+			prefix='\n'.join(day.split('\n')[day.split('\n').index(
+				'exit 0')+1:])+'\n';continue
+		elif day.split('\n')[0][:4] == '<!--': suffix=day;continue
+		with open(day.split('\n')[0]+'.html', 'x') as g:
+			g.write(prefix+day+'\n'+suffix)
+";cd blah;for f in *.html;do #in glob we trust
+test -z "$last" || sed -i "s,__NAVIGATION__,$nav<A HREF=\"$f\">\&gt;</A></P>," \
+"$last";nav="<P>";test -z "$last"||nav="$nav<A HREF=\"$last\">\&lt;</A>"
+nav="$nav<A HREF=\"index.html\">^</A>";last="$f";done
+sed -i "s,__NAVIGATION__,$nav</P>," "$last";for f in *.html;do #e unibus puellam
+fi="$(echo "$f" | cut -d . -f 1)";test "$fi" = "index" && continue
+printf '<A HREF="/blah/%s.html">%s</A>\n' "$fi" "$fi"; done|sort -r|\
+sed -e "1i<!DOCTYPE html><HTML><HEAD><TITLE>blah</TITLE></HEAD><BODY><PRE>\
+<A HREF="..">..</A>" -e '$a</PRE></BODY></HTML>'>index.html
+exit 0
+
+	That's the source code to this blog, in its entirety. My writing
+process was simple:
+	- write the beginning and initial Python portion
+	- pass out
+	- wake up at 0600 not knowing who or where I am
+	- see this code and continue it
+	- pass out again
+	- wake up at 1700 knowing who but not where I am
+	- write most of the rest
+	- pass out again
+	- wake up half an hour later, finish
+
+	It's organized in sections though it doesn't appear to be organized
+whatsoever:
+
+#!/bin/sh
+set -ex
+mkdir -p blah
+python -c "
+import os
+os.chdir('blah')
+with open('../$0', 'r') as f:
+	for day in f.read().split('\n\n\n'):
+		if day.split('\n')[0] == '#!/bin/sh':
+			prefix = '\n'.join(
+				day.split('\n')[
+					day.split('\n').index('exit 0')+1:
+				
+				]
+			) + '\n'
+			continue
+		elif day.split('\n')[0][:4] == '<!--':
+			suffix = day
+			continue
+		with open(day.split('\n')[0]+'.html', 'x') as g:
+			g.write(prefix + day + '\n' + suffix)
+"
+
+	This splits the blog into days, where each day is delimited by three
+newlines. Every day is two lines apart.
+	A day that starts with the POSIX shell shebang is the /prefix/, which
+is prepended to each day. It cuts off everything until after "exit 0", the end
+of the script, and after that is the actual HTML prefix to each blah page.
+	A day that starts as an HTML comment is the /suffix/, appended to each
+day. This obligates an HTML comment at the end of each post, the same comment,
+so I just made it something sort of interesting yet sort of bog standard.
+	I explained this poorly but I spread the code out so it's a little
+easier to read, I think it's pretty simple. git.sr.ht/~trinity/homepage/tree
+/main/blog, you can see how it's laid out.
+	Each day, prefixed and suffixed, is output as its own [day].html to the
+created blah/ directory.
+
+	Next:
+
+cd blah
+for f in *.html
+	do
+		test -z "$last" || sed -i \
+-e "s,_NAVIGATION_,$nav<A HREF=\"$f\">\&gt;</A></P>," "$last"
+		nav="<P>"
+		test -z "$last" \
+			|| nav="$nav<A HREF=\"$last\">\&lt;</A>"
+		nav="$nav<A HREF=\"index.html\">^</A>"
+		last="$f"
+	done
+sed -i "s,_NAVIGATION_,$nav</P>," "$last"
+
+	This replaces _NAVIGATION_ with an actual navigation bar. The actual
+string has two underscores before and after NAVIGATION but this blog is held
+together with shoelaces and bubble gum and I don't wanna fuck around and find
+out.
+	I don't know how this works, I let my fingers handle the flow.
+		(The secret is that I just run it in my head and adjust the
+		 beginnings and ends until it runs in my head for two times
+		 correctly. Then as long as state doesn't drift it's all good.
+		 This is fucky and I don't know how to explain it and I don't
+		 really know how it all goes about but you can do really
+		 complex but really really tight program flow just by vibing
+		 against it and letting your fingers tap tap tap, yknow?)
+
+Next:
+
+for f in *.html
+	do
+		fi="$(echo "$f" | cut -d . -f 1)"
+		test "$fi" = "index" \
+			&& continue
+		printf '<A HREF="/blah/%s.html">%s</A>\n' "$fi" "$fi"
+	done \
+		| sort -r \
+		| sed \
+			-e "1i\
+<!DOCTYPE html><HTML><HEAD><TITLE>blah</TITLE></HEAD><BODY><PRE><A HREF="..">..</A>" \
+			-e '$a</PRE></BODY></HTML>' \
+		> index.html
+exit 0
+
+	This takes all the files in blah/, builds an index, adds a prefix and
+suffix to the stream, and outputs it all to blah/index.html in one go. This is
+the simplest part of the script and I was worried it would be hard but it
+wasn't really, it just required a little bit of embracing of UNIX piping.
+
+	["Streambreak"]: After experiencing a genocide, Ada Karina time travels
+back to the past to prevent it from happening. However things start diverging
+from plan when a soup-fueled arsonist grows from nuisance to idol to
+geopolitical disaster.
+	["Antero"]: Tales from a future dystopia where the very formation of
+memories is outlawed.
+	["Sponge"]: Olive Edgerton is an employee at an impossibly popular
+burger joint, where every ingredient is grown or produced in-house.
+	["Saikokon"]: After an apocalypse, the last survivor is selected as an
+exhibit at Saikokon, a conference for psychic time travelers.
+	["Pasture"]: Tales from after the end of the world.
+
+
 2023-02-06
 
 2022年03月02日
