Design of Tumble Manager

TumbleManager Runs for the first time, or for a total site build.

1. Figures out where the source directory is.

2. Figures out where the publication directory is.

3. Copies over any files that don't need processing (based on file name?)

4. Captures items for publication

   • collect meta data (post time, title, tags, author)

   • create index of posts to be (re)generated.

   • produce site-wide indices. (e.g. archive). place anchor at top for future additions to the file.

   • copy generated files to production directories

5. Page Generation

   • apply content into template content-atom

   • apply atom-templates into full page templates

   • if in latest n posts, insert atom into index page

   • copy production files to production directories.

TumbleManager runs subsequently

1. sort "to be published" posts into a stack of some sort (this function is probably plug able/advisable by design), and pulls the top 1 post in for publication.

2. Captures item for publication

   • collect meta data for post (post time, title, tags, author)

   • insert link at anchor in the archive index

   • copy generated files to production directories

3. Page Generation

   • apply content into template content-atom

   • apply atom-templates into full page templates

   • insert atom into index page

   • remove an item from the bottom of the index page.

   • copy production files to production directories.

Publish Changes

(some sort of facility to be able to edit previously published posts without rebuilding everything.) Possibly:

• set a revised value to 0 on all posts by default. (not need to be specified.)

• scan all published files for a non-zero value in this field.

• run page generation functionality. links become immutable. posts on the index aren't changed.

• call it good.

Alternative is to hash everything, every time. not a huge loss, but annoying.