Documentation of rss patch

3 files changed, 27 insertions, 0 deletions

diff --git a/lists/dirty-haskell/002 b/lists/dirty-haskell/002
new file mode 120000
index 0000000..ed24126
--- /dev/null
+++ b/lists/dirty-haskell/002
@@ -0,0 +1 @@
+../../posts/blog-rss.md
\ No newline at end of file
diff --git a/lists/zz_all/002 b/lists/zz_all/002
new file mode 120000
index 0000000..ed24126
--- /dev/null
+++ b/lists/zz_all/002
@@ -0,0 +1 @@
+../../posts/blog-rss.md
\ No newline at end of file
diff --git a/posts/blog-rss.md b/posts/blog-rss.md
new file mode 100644
index 0000000..559a6c6
--- /dev/null
+++ b/posts/blog-rss.md
@@ -0,0 +1,25 @@
+% dirty-haskell.org´s rss feeds
+
+I extended the software suite inherited from [math.kleen.org](http://math.kleen.org) to include support for rss feeds.
+The heart of the issue is a ~80 line haskell script I chose to call, in a bout of creativity, "generate-rss.hs".
+The script uses the [feed](http://hackage.haskell.org/package/feed-0.3.9.2) package.
+
+generate-rss.hs gets passed a title and a list of paths below ./lists to incorporate as items.
+It generates an empty feed structure, adds title and a (hardcoded) base url for RSS metadata, and iterates over the given paths — generating for each path an item to be included in the finished feed.
+This procedure makes use of a state monad (StateT (Feed, Maybe ClockTime) IO ()) to sequentially add items to the feed and keep track of the modification/change time of the newest path examined.
+Each item carries a title, an url, a date, and contents as follows:
+
+-  The date used is the modification/change time of the path supplied as a command line argument at the beginning of the program (usually a symbolic link in ./lists) — as such it is the time the post was linked into the particular list we´re generating a RSS feed for (this was not a deliberate design choice but a side effect of the canonical implementation — it was later decided that this behaviour was in fact the one expected all along).
+-  The url is generated by following, recursively, the trail of symbolic links starting in ./lists, assuming the final target is indeed in ./posts, and forming the filename of that target into a (hopefully) functional url in a hardcoded fashion.
+-  The title is extracted from the markdown file using a function shamelessly copied from extract-title.hs (The author wrote that one too, after all).
+-  The contents are read into Pandoc and rendered into [AsciiDoc](http://en.wikipedia.org/wiki/AsciiDoc) format (it seemed convenient at the time).
+
+Along the way two helper functions were introduced — if an implementation of those already exists in Prelude or somewhere else common please mail in a comment:
+
+    (<->) :: [(a -> b)] -> a -> [b]
+    [] <-> _ = []
+    (f:fs) <-> x = (f x:fs <-> x)
+
+    (<-->) :: [(a -> a)] -> a -> a
+    [] <--> x = x
+    (f:fs) <--> x = fs <--> (f x)