HTML style rules

[A. Pagaltzis]

* Kragen Sitaker <kragen at pobox.com> [2002-01-16 09:38]:
> Tools I'd like:
> 
> - a linkrot preventer that keeps track of my outgoing links,
>   checks them periodically to see if the pages they link to are
>   still there and still more or less the same, and *quickly*
>   makes my links to them either disappear or go to a local copy
>   of the page if not, based on a predefined policy
> - something integrated with the linkrot preventer to help me
>   with link titles
> - something to help keep width and height attributes up to date
> - help from the linkrot preventer to predict linkrot by URL:
>   have similar URLs (e.g. same domain name, nearby IP
>   addresses, same URL structure) rotted quickly in the past?
>   Are current pages on the site all new, or are there some old
>   ones?

Interesting thoughts. I’ll skip on the predictive heuristics as
I have an aversion to such machinery; also, I won’t make my link
checker actually edit my pages for me. Finally, keep in mind that
what my complete weblog (and in the fullness of time, the
entirety of my site) lives as XHTML inside a single well-formed
XML file. That all said, here are the notes I just added to the
entry for my weblog in my TODO file:

> * Links:
>   * Authoring:
>     * make use of the `type` attribute; store target MIME type
>     * if the link is to `text/*`, put a namespaced attribute in
>       the link with a snippet quoted from the page
>   * Checking: links are…
>       * … OK if they are 2xx, have the recorded MIME type (but
>         HTML == XHTML), and contain the quotation (if given)
>       * … in need of rewrite if they are 30[12] but the rest of
>         the checks pass
>       * … in need of manual intervention otherwise
>   * Further possibilities: might be useful to preserve the
>     `Last-Modified` and `ETag` headers in more attributes

Regards,
-- 
Aristotle Pagaltzis // <http://plasmasturm.org/>