diff --git a/README b/README
index b9e9121..9bc1022 100644
--- a/README
+++ b/README
@@ -1,24 +1,33 @@
-Software prerequisites: apt-cacher-ng debiman mandoc python3
+Software prerequisites:
+    - apt-cacher-ng
+    - debiman (install from repository, follow instructions in README)
+    - mandoc
+    - python3
+    - python3-venv
 
 After installing apt-cacher-ng, the proxy is started automatically and should be
 listening on port 3142.
 debiman by default will download from http://localhost:3142/deb.debian.org/
-This setting can be changed with -local_mirror or -remote_mirror
+This option can be changed with -local_mirror or -remote_mirror. Non-LTS Debians are
+removed from the mirrors; use the archive instead: https://archive.debian.org
 
 How to use the scripts:
 
 1. Download manpages from Debian:
 
-       $ DEBIMAN_SERVING_DIR=/absolue/path make download
+    $ MIRROR="https://archive.debian.org" CODENAME="buster" DEBIMAN_SERVING_DIR=/path make download
 
 2. Extract all the pages from .gz:
 
-       $ DEBIMAN_SERVING_DIR=/absolue/path make extract
+    $ DEBIMAN_SERVING_DIR=/path make extract
 
 3. Convert manpages to plaintext and html:
 
-       $ DEBIMAN_SERVING_DIR=/absolue/path make convert
+    $ DEBIMAN_SERVING_DIR=/path make convert
 
 4. Create RDF graph:
 
-       $ DEBIMAN_SERVING_DIR=/absolue/path ./rdf.py
+    $ python3 -m venv venv
+    $ source venv/bin/activate
+    $ pip install rdflib
+    $ CODENAME="buster" DEBIMAN_SERVING_DIR=/path ./rdf.py
diff --git a/scripts/Makefile b/scripts/Makefile
index 3d83416..cc4485f 100755
--- a/scripts/Makefile
+++ b/scripts/Makefile
@@ -1,31 +1,35 @@
 SHELL = /bin/bash
 
-# THIS PATH MUST BE ABSOLUTE.
+# The mirror to download from
+MIRROR ?=
+
 # The path where debiman will save files.
-# There is a bug in debiman, it won't accept a relative path
-DEBIMAN_SERVING_DIR ?= ./
+DEBIMAN_SERVING_DIR ?=
 
+# Name of the distribution to be processed
+CODENAME ?=
 
 # This just prints out variables for displaying
 vars:
-	@echo "Using envvar DEBIMAN_SERVING_DIR = ${DEBIMAN_SERVING_DIR}"
-	@echo "If you get errors, this path MUST be absolute."
+	@echo "MIRROR=${MIRROR}"
+	@echo "CODENAME=${CODENAME}"
+	@echo "DEBIMAN_SERVING_DIR=${DEBIMAN_SERVING_DIR}"
 
 # Note: the behaviour of debiman is to download all the manpages, then render them all.
 # This cannot be changed. Since we're only interested in the raw manpages and not
 # the HTML output, -only_render_pkgs is a hack that will make debiman render only one
 # page (0ad) and quit.
 download: vars
-	debiman -sync_codenames="buster,bullseye,bookworm" -sync_suites= -serving_dir="${DEBIMAN_SERVING_DIR}" -only_render_pkgs="0ad"
+	debiman -remote_mirror="${MIRROR}" -sync_codenames="${CODENAME}" -sync_suites= -serving_dir="${DEBIMAN_SERVING_DIR}" -only_render_pkgs="0ad"
 
 # Extract downloaded pages since they're compressed by default
 extract: vars
 	find "${DEBIMAN_SERVING_DIR}" -type f,l -name "*.gz" -exec gunzip --decompress --force --keep "{}" \;
 
-# Convert manpages from roff to plaintext
+# Convert manpages from roff to other formats
 # Manpage files are named "page.section.lang".
 # TODO ! -name "stress-ng.1.en" ! -name "md.4.en"
-#      this is a hack for skipping the rendering of those pages. The version of mandoc
+#      this is a hack for skipping the parsing of those pages. The version of mandoc
 #      in Debian is outdated and gets stuck in a infinite loop. Remove this hack if
 #      using a more recent mandoc.
 convert: vars
diff --git a/scripts/rdf.py b/scripts/rdf.py
index f31daf9..3db52db 100755
--- a/scripts/rdf.py
+++ b/scripts/rdf.py
@@ -8,8 +8,13 @@ import sys
 from rdflib import BNode, Graph, Literal, Namespace, URIRef
 from rdflib.namespace import RDF
 
+CODENAME = os.getenv('CODENAME')
 DEBIMAN_SERVING_DIR = os.getenv('DEBIMAN_SERVING_DIR')
 
+if not CODENAME:
+    print('envvar CODENAME is not defined')
+    exit()
+
 if not DEBIMAN_SERVING_DIR:
     print('envvar DEBIMAN_SERVING_DIR is not defined')
     exit()
@@ -34,7 +39,7 @@ def percent_encode(string):
                  .replace('[', '%5B') \
                  .replace(']', '%5D')
 
-for absolute_file_path in pathlib.Path(DEBIMAN_SERVING_DIR).glob('**/*.roff'):
+for absolute_file_path in pathlib.Path(DEBIMAN_SERVING_DIR, CODENAME.lower()).glob('**/*.roff'):
     if not absolute_file_path.is_file():
         exit('Not a file: {}'.format(absolute_file_path))
 
@@ -73,9 +78,8 @@ for absolute_file_path in pathlib.Path(DEBIMAN_SERVING_DIR).glob('**/*.roff'):
     except:
         html = ''
 
-    # Create a separate graph node for this manpage
-    # A separate, temporary graph for a manpage that is printed out immediately
-    # because I haven't got enough RAM for storing thousands of pages.
+    # Create a new graph for this manpage that is printed out immediately
+    # because I haven't got enough RAM for storing the whole graph.
     g_page = Graph()
 
     page_ref = URIRef(percent_encode(f'dokk:manpages:debian/{distro_number}/{distro_package}/{filename}'))
@@ -94,6 +98,9 @@ for absolute_file_path in pathlib.Path(DEBIMAN_SERVING_DIR).glob('**/*.roff'):
 
     print(g_page.serialize(format='nt'))
 
+    # Now we're going to create nodes for the debian distro as well as the package
+    # this page belongs to.
+
     # Create a graph node for this package
     # Link to the page node
     package_ref = URIRef(percent_encode(f'dokk:manpages:debian/{distro_number}/{distro_package}'))