Evan Williams: Today, I added better permalinks
Yesterday, all my relative links worked. And all content was served with the correct file type.
Evan Williams: Today, I added better permalinks
Yesterday, all my relative links worked. And all content was served with the correct file type.
Today, I have better permalinks, but I'm still squashing bugs. When I am done, relative links, mime types, old-style and new-style URLs will all work, but I fear that I am not there yet.
Some day in the near future, I should be able to restructure my internal storage (optimizing for new-style permalinks vs the old style); and set permalink names separate from titles.
One side effect of this change: searches which return only one result no longer show only the excerpt for that one result; now you see the full entry and all the comments for that entry.
Anne - which feed?
At the moment, <guid>
and rdf:about
are unchanged (i.e., they still have the old URI structure). <link>
has the new structure.
Suggestions welcome.
Sam,
Excellent URLs! I recently re-wrote my blog app to work in just this way. So, not only are the URLs logical and descriptive, but they can be truncated in meaningful ways (all posts for March, 13th 2004, all posts for May, 2003, etc).
I chose to forgo the directory semantics (/yyyy/mm/dd/index.rss) in favor of /yyyy/mm/dd.xml, but I think the end result is the same.
Ah, I see. Very nice. When I did my code re-write, I had to support all kinds of legacy URLs (damn Google!). Normally this would have been easy with mod_rewrite, but since the Tomcat plug-in (mod_jk) doesn't pass the original URL through the pipe, all the code I have that keys off the requested URL couldn't function. Which pushed tons of regex into Java code (!), which sucks.
Keep up the good work.
Why do everyone else's excerpts seem to make sense, and mine are always strange, barely related things, half the time from nine months ago? (Okay, I know why that is, I changed the order of autodiscovery on you, so now you're getting a different feed.)
Suggestions? My preference would be either for both link and guid, pointing to the same thing, or if we are certain that there aren't any readers which will use a URI from guid but won't use one from link, then permalink in link and an isPermalink="false" guid in guid, if the thought of the duplicate URIs would make you less happy about your feed. The two different URIs were interesting, in that I hadn't quite noticed that Bloglines will link guid from the entry title and link from the word "Link" if they are both present and different, but I'd rather not relearn that every time you post. However, that's the preference of someone who doesn't have to write the special-case code to not change the guid for existing entries, while giving a different sort to future entries.
And Phil, you know as well as I do, the reason for the delay in this excerpt appearing was because I was literally laying down the tracks for support for the new style URLs while the trains (in this case, referers) were still coming. Thanks for providing a live test case!
Oh, and repeat after me: isPermaLink, isPermaLink, isPermLink, with a capital eL.
For now, I've gone back and changed my rss2 feed to indicate that my guids are not my preferred permalinks.
Oh, I didn't mind the delay at all (I would have just come over here and commented, if I was in a hurry). I just meant that I'm seeing excerpts show up as new from things we said in my comments last summer or earlier, I think because when you check the referrer, you now find the per-entry RSS feed and get the link for the particular comment. No big deal, it just makes me look a little spammy when I pop up in your comment feed with an old stale link.
L L L. LL. He. At least I've learned that the default value when it's missing is true ;)