X-Git-Url: https://plomlompom.com/repos/?p=redo-blog;a=blobdiff_plain;f=README.md;h=49ad8b182217cb3dcf988d7218871b18144d09f8;hp=9d6a05b17e7cbffe92d6f62d90315fa5f8724102;hb=ed5ca83e98679f5f3ce50392320b427a4ac11e02;hpb=6b5631096b03f1c985bc40c4d2e244315280c686

diff --git a/README.md b/README.md
index 9d6a05b..49ad8b1 100644
--- a/README.md
+++ b/README.md
@@ -1,14 +1,15 @@
 redo-blog
 =========
 
-small blog system using the redo build system
+small blog system using the redo build system, with blog article files written
+in (pandoc) Markdown or ReStructured Text.
 
 dependencies
 ------------
 
 - redo
 - python3
-- uuidgen
+- uuidgen (Debian package: uuid-runtime)
 - html2text
 - pandoc
 
@@ -28,13 +29,79 @@ You can then enter the directory and run redo there. This will generate article
 These files will be linked to symbolically in a directory ./public/.
 
 Some metadata files will also be generated below ./metadata/: For each article,
-there will be generated a .uuid and a .intermediate file; furthermore, files for
-data used in ./feed.xml and ./index.html will be built there and can be edited
-to customize the blog – namely the files url, author, uuid, title, index.tmpl,
-index_snippet.tmpl, article.tmpl.
+there will be generated a .automatic_metadata (to contain an article's UUID,
+checksum, and creation/modification dates) and a .intermediate file (to contain
+pandoc-formatted article content like title and body); furthermore, files for
+data used in ./feed.xml and ./index.html will, if non-existant, be built there
+and can be edited to customize the blog – namely the files url, author, title,
+index.tmpl, index_snippet.tmpl, article.tmpl. A blog-specific UUID and creation
+date is stored in ./metadata/automatic_metadata
 
-bugs
-----
+recipe to remotely manage a redo blog with git
+----------------------------------------------
+
+On your server, install the dependencies listed above. Then set up a repository
+for your blog files. Let's assume we want it to sit in our home directory and be
+called `blog`:
+
+    cd ~
+    mkdir blog
+    git init --bare blog.git
+    cat << EOF > blog.git/hooks/post-update
+    #!/bin/sh
+    blog_dir=~/blog
+    export GIT_DIR=\$(pwd)
+    export GIT_WORK_TREE="\$blog_dir"
+    git checkout -f
+    cd "\$GIT_WORK_TREE"
+    redo
+    git add metadata/author metadata/url metadata/title metadata/*.tmpl metadata/*.automatic_metadata metadata/automatic_metadata
+    status=\$(git status -s)
+    n_updates=\$(printf "$status" | grep -vE '^\?\?' | wc -l)
+    if [ "\$n_updates" -gt 0 ]; then
+      git commit -a -m 'Update metadata'
+    fi
+    EOF
+    chmod a+x blog.git/hooks/post-update
+
+Enable management of `~/blog` via redo-blog:
+
+    git clone https://github.com/plomlompom/redo-blog/
+    cd redo-blog/
+    ./add_dir.sh ~/blog
+    mkdir ~/blog/public
+
+Link to the `public` subdirectory from wherever your web server expects your
+public web content to sit:
+
+    ln -s ~/blog/public /var/www/html/blog
+
+Client-side, do this (you obviously need to customize this code; at least
+replace the username `user` and the server name `example.org`):
+
+    cd ~
+    git init blog
+    cd blog
+    git remote add origin ssh://user@example.org:/home/user/blog.git
+    mkdir metadata
+    echo 'https://example.org/blog/' > metadata/url
+    git add metadata/url
+    git commit -m 'set up blog metadata'
+    git push origin master
+
+If successful, the git hook will furthermore commit some ~/blog/metadata/ files
+generated by redo, that can be pulled into the client-side local repository:
+
+    git pull origin master
+
+bugs and peculiarities
+----------------------
 
 Don't create a index.rst or index.md in the redo-managed directory, that will
 break things.
+
+The article title is derived in .md files from a first line prefixed with `%`,
+while all other headings are treated as sub-headings. In .rst files, the title
+is derived from a heading that must be at the top of the document, and be of an
+adornment style (such as underlining with `=`) that must be used only once in
+it.