... something borrowed, something blue.
What started me down this path was posts by Tom Coates and Phil Ringnalda. Together, they started me thinking about rethinking my original story vs. post dichotomy, and toward what is emerging as the blog standard of post vs. remaindered links dichotomy.
I've also updated a number of my feeds - notably, I've updated the rss 1.0 and rss 2.0 feeds to conform to description being an optional synopsis. My RSS 2.0 feed has replaced link with guid - I think all aggregators that have RSS 2.0 support can handle this, but if there are any that have lagged behind, I still have plenty of other feed formats to choose from. Finally, I've removed a number of usages of admin and dublin core namespaces from the rss 2.0 feed (I still use dc:creator in my comment feed to avoid revealing email addresses).
While the header is shamelessly borrowed from Bob Congdon, I plan on evolving it.
I've tested all this under a number of browsers and aggregators, and I don't believe I have broken anything, but if I have, let me know.
The fonts on the new design look awful under ClearType on Windows XP - MS Sans Serif does not get smoothed like... well, just about every other font. Any way you can insist on another font?
I'm only a relatively new reader but... is there an Atom feed for this blog? I haven't found one yet.
Sam,
Thanks for removing the usage of the admin-prefixed namespace from your RSS feed. The next version of RSS Bandit exposes the value of the webMaster and managingEditor elements to users when errors occur. I was debating on adding support for the admin-prefixed namespace because of your feed since it didn't use the standard elements provided for this in RSS 2.0. With your recent decision I won't worry about supporting such funkiness.
PS: I dislike the new design but it isn't like I actually visit your website on a regular basis so more power to you.
Neil: I've switched back to "Helvetica, sans-serif", any better? I do have an atom.feed. At the moment, it uses a relative URL for links, something that looks like it will be adopted in the next rev of the spec.
Dare, the admin namespace predates RSS 2.0 by quite a bit, and is still widely used. I didn't have to get past the B's in your blogroll to find this feed. In any case, that namespace is still used in my RSS 1.0 feed.
Email addresses are optional on my weblog, and if a web site is provided, it trumps the email address for display and syndication purposes. So, the overwhelming majority of the entries in my comment feed do not reveal email addresses.
David, xhtml:body elements are on quite a few weblogs and are widely supported. Dare's feed and aggregator are examples.
When I clicked on this link from my aggregator, I noticed your updated site design, and I wanted to offer a compliment. Great work!
Back on topic, the feed I'm using works perfectly in Bloglines.
Obviously trying to make sense of the Userland RSS specs is a fool's errand, but my reading of the spec is that <link> is the click-through link (which can point anywhere the publisher wants to, as far as I can tell). <guid> is defined as "A string that uniquely identifies the item". The permalink functionality seems like an afterthought on top of that.
Dare: Shrook 1.x was written under this assumption (and the few feeds I'd seen with guids at the time) and thus <guid> is used only as an identifier. Shrook 2 uses it as the click-through link when <link> is absent, since this is the way people have ended up using it.
Mark: <guid> is the permalink. <link>, as always, may or may not also be.
Graham,
I find the Userland spec to be quite clear in this regard. The guid is pretty much equivalent to URI as we know them today (unique identifiers that might lead to network retrievable resources). The Userland spec then specifies that a guid can indicate that it is a permalink either by having no isPermaLink attribute or having one whose value is "true". It is too bad that the W3C/IETF/etc did not see fit to build this functionality into URIs (which we did have in the days of URNs and URLs but I digress...)
The link element may or may not be a permalink but does point something relevant to the content or title (if there is no content). There is no requirement for a link to be a permalink, but in the absence of a guid it's the next best thing. This doesn't necessarily mean a link should be assumed to be the permalink, just take a look at [link] to see an example of where this is not the case.
The simples thing that could possibly work would be to provide a link element.
Are there aggregators that do not support guid as a permalink? How do they read this feed? [link]
I like this design a lot better than the old one. It's cleaner, with less distracting use of sidebars and colours.
My only real complaint is with the rounded corners: First off, they only render in Mozilla, and Mozilla doesn't do a very good job of rounding them - some anti-aliasing would be appreciated.
I presume that you are using the Mozilla properties because you don't want style to influence your markup. I have created graphical rounded corners with no impact on markup that work in both Mozilla, Safari and Opera. (The site is down at the time of writing, but I expect it to be back up in reasonable time - an alternate version, using shadows as well, exists in a a w3c tutorial
Arve, my preferences are for a site which is image free and scalable - to me the ideal would be something like SVG. However, I realize that this is not practical at this point in time.
My problem with tutorials such as the one that you pointed me to is that they feel like some of the cooking shows you might see where the main ingredients were prepared ahead of time and the cook makes a big deal about putting it in the oven. That's nice, but what if I don't like light green boxes? Are there readily available and tweakable instructions for generating the images used? What I would prefer is a standard textual representation that I can edit and then pass through a tool that generates the icon. Again, SVG comes to mind.
Thanks in advance for any pointers.
Sam, My own tutorial is now back up. If you study that, you'll notice that you don't need to know the background color of the element you're styling - you instead need to know the background colour of the element it is being drawn on top of (In other words, you need to know the background colour of the body). Neither does this technique cost much in terms of "page weight" - the images in this tutorial are in all 1962 bytes. If you drop the decoration, they would drop by another 600 bytes, which should load in less than a second on a 28,800 connection. And yes: it is scalable. (The w3c tutorial is not too scalable, though)
I am myself using a variant on this technique in my other blog to create drop shadows that scale to pages of any width.
While my tutorial doesn't cover graceful degradation in itself, making that happen is easy, while keeping CSS valid.
Arve's rounded corner tutorial is just brilliant, Sam. Have a look at it. Even if you don't decide to implement it, it's stuffed with "neat to know" stuff. And to your new design; I like it. I feel that the search box is floating a bit unattached "in space", but other than that, I think it's good.
After all, it is as you say; It's all data. (Hence, presentation and visual googlygook is secondary.)
PS: Your posting engine doesn't like left or right pointing double angle quotation marks (unicode U+008B and U+00BB). It screams the following:
CGI Failure
traceback:Traceback (most recent call last):
File "gateway.cgi", line 39, in ?
post()
File "/home/rubys/mombo/post.py", line 193, in post
body=html2xml(spellcheck(body))
File "/home/rubys/mombo/spellcheck.py", line 14, in spellcheck
tmp.write(text.encode('iso-8859-1').replace(u'’',"'"))
UnicodeError: ASCII decoding error: ordinal not in range(128)
Testing: ‹ « O » ›
OK, looks fixed now.
«Great»! :-)
But, I'm having problems with your new URI scheme, Sam. Opera 7.5TP3 doesn't understand it, possibly because they are relative(?). No big problem, but I thought you should know. This is the feed i subscribe to.
Lastly: Why can't I put 'title' attributes in the <a> tag?
Sam, I was going to ask a question about where in Opera Asbjoern was seeing problems, and got this error:
traceback:Traceback (most recent call last):
File "gateway.cgi", line 39, in ?
post()
File "/home/rubys/mombo/post.py", line 197, in post
body=html2xml(spellcheck(body))
File "/home/rubys/mombo/spellcheck.py", line 14, in spellcheck
tmp.write(text.encode('iso-8859-1').replace(u'’',"'"))
UnicodeError: ASCII decoding error: ordinal not in range(128)
(I presume the error was in the newsreader)
Test: Asbjørn Ulsberg. OK, another bug fixed. This fix is better (i.e., more comprehensive).
I also would be interested in that question. To my knowledge, I did not change the url scheme.
I have taken a look at the tutorial. Good stuff! My biggest concern is how to generate those images. Perhaps once I get back from SXSW, I'll download batik and play with SVG again.