I think weblog URLs should all have a date and a slug to identify their content. For instance, the URL for this post is http://keithdevens.com/weblog/archive/2004/Jan/06/permalinks
Now, weblog URLs should always have the date in the URL. But in the absence of a slug, what's are good ways to uniquely identify a post? Possibilities include:
- A timestamp of the creation time, in hour/minute/second format
- A timestamp of the creation time, in Unix timestamp format
- An arbitrary unique ID for the post (usually just the primary key of whatever table stores your weblog entries)
- A GUID (some news sites use this for article IDs)
- A number indicating the order the post was posted in the day (first post is 1, etc.) I suppose you could do a textual representation of this as well ("one","two","forty-seven")
- Some serialization of the title, if one exists
- What else?
In my home grown system we decided to go with year/month/date originally (hey, it was back in the 1990's) and we later updated to doing year/month/date/hour/min/secs, so a month gets this '200401.html' and a day gets this '200401.html#01072004' and an entry gets this '200401.html#01072004072340'
It's not the cleanest, but I do prefer to use the date/time than some arbitrary id number assigned by a system I might migrate away from someday...