A little plug here for my friend Jeff. He's been working on an
email utility called mailpie, a set
of command-line programs providing full-text indexing and searching of
large email archives. I'm not using it yet, because I've been too lazy
to set it up, but it looks pretty cool.
Lately, Jeff and I were discussing the limitations of our approaches
to reading our feeds. During a fit of outsourcing, I settled on Google Reader. Jeff continues to
prefer to make it tough for advertisers to mine his data, opting for Planet. Like any software,
both have minor issues.
I like Google Reader's keyboardable interface. That makes old
Unix-heads like me happy. I don't like that old, unread entries don't
quietly slip away. There's a lot of pressure from the line on the
navigation pane that says, Liberal Rubbish (1000+)
. It makes me
feel bad for not keeping up, and makes reading my feeds seem like work.
Probably this could be solved with a Mark as read after some period
of time
setting, but that currently doesn't exist. I'd like
them to expire based on the time-to-live specified by the feed itself.
Feeds are by their nature ephemeral, and keeping them around and all up
in my fries misses the point. When I look at my feeds, I want to see
what's new, not the history of the website back to the beginning of
time.
I also feel a little dirty using Google Reader because I know that
all my clicks, choice of subscriptions, and entry keywords are being
inspected and dissected by Google's army of virtual demographers and
adaptive hindbrain delivery scripts, so my brain can be picked apart and
sold to advertisers. Nevermind that my browser provides anti-marketing countermeasures;
knowing this is going on is irksome.
Yes, blah blah business model. Yes, blah don't be evil
blah. Yes, I know my data is not linked back to me
individually. I know, I know. I love the power the Google gives
me, and I take advantage of it. It just squicks me out sometimes.
Bonus points for the Google's brilliant, intutive, usable AJAX
interface, though.
My primary reason for not choosing Planet is essentially a bug: the
feeds are cached, and the caches are never cleaned-up, growing to fill
the disk space they have available. Lacking the desire to worry about
solving this problem, along with going through an offloading phase,
during which I started using flickr seriously, along with Google
Calendar and Reader, steered me away from Planet.
...So Jeff and I were talking about feeds and readers the other day.
Because we both choose mutt as our
email clients, with appreciation for its power and flexibility, Jeff
said, I should write something like mailpie for feeds. It would put
together mbox mailboxes, readable from mutt.
As I've been
working on the next generation of my feed sidebar-plugins for pyblosxom, I've been
thinking about that idea. Here are a few of the things that have crossed
my mind.
Pros:
- Mutt has a well-known and convenient user interface.
- Like email messages, feed items are discrete chunks.
- There are no distracting bells and whistles in mutt. Just cool,
comfortable text.
Cons:
- Feed items often have embedded links the reader would like to
follow. Mutt does not support links directly (or any rendering of
embedded html, for that matter). The user would have to contrive to
run a text-based browser to display each feed entry, or strip the
formatting entirely and devise a way to display links so the user
could take advantage of any terminal-based utilities for following a
link.
- Feed items, while not dependent on formatting, may be
multimedia. YouTube videos do not fit well in mutt. You'd have to
fire up a browser to see them. (Or embedded pictures or podcasts
or...)
- Unlike email messages, feed items are typically not threaded.
One of the major benefits of mutt is irrelevant.
- Barring some kind of intelligent downloading scheme, feeds that
are not full text are inconvenient to read if you're pretending
they're something like email messages. Of course, they're
inconvenient to read, whatever reader is used.
It’s not exactly a
pot of gold at the end of a rainbow, but Lincoln Mayor Chris
Beutler is considering dipping into ... (5 comments)
Thanks,
journalstar.com.
- Mailpie's strength lay in its full text search of an archive of
email messages. I don't want an archive of feed items. The web is
essentially that, using Google as the index.
The jury's still out. The idea of a text-mode, minimally formatted
feed reader is attractive. (Any of you whippersnappers remember Gopher?).
The question is how to present the potentially rich set of data with
links usefully. Maybe its just a matter of providing a properly
configured mutt...
At the time I'm writing this, the design of my site is minimal. I've
deliberately chosen xhtml 1.0
strict as the markup for my pages, and there are currently exactly
four lines of css, the
purpose of which is to present the tags for each entry on a single line.
In markup, they are noted as an unordered list, which, semantically,
tags are.
Philosophically, my approach is to minimize the use of styles,
relying on them only to make the page more readable, while maximizing use of
semantically significant markup, such as an on ordered list of tags, instead of
whitespace-separated, inline links. It's kind of like CSS Naked Day, but it's
everyday.
This gives control back to the user where it should be, but also
requires participation and a degree of savvy that users may not be used
to. Users are given responsibility for using—users using,
hmmm—their browsers to find a font size they find comfortable, for
example. To a greater extent, users must take responsibility for their own experience.
It's similarly liberating for me as generator of the content. I'm not
a graphic artist, and I'm certainly not interested it tweaking and
retweaking the layout to make it work with ahem other browsers.
I want to be reasonably confident that I've made the information
available in the most convenient and easy to digest form, and then not
worry about it. I have a gut feeling that if ease of use is
accomplished, beauty of presentation will be a co-effect.
Unfortunately, we have been trained by corporate marketing experts to
expect data to be presented in a highly controlled and prepackaged
fashion. Content that is not presented in three non-resizable columns is
likely to be written off as unprofessional
or unreliable
.
We have been trained that real information is branded.
I've received mixed reviews so far, ranging from very clean
to
unpolished
to it renders so fast
to 1996 called; they
want their layout back
. I'll keep adjusting as
old features are reapplied and new features are added, but I don't
expect it will change much.
While tinkering around with the stylesheet for the pyblosxom implementation
of Internal Dialogue, I came across something curious.
-
My first crack at applying a background color (so I wasn't looking at
a blinding sea of white) was pretty simple.
body {
background: #cc9;
}
But it didn't quite look right, as you can see below.
-
So I tried fiddling with the margin...
body {
background: #cc9;
margin: 0px;
}
-
To try to get some idea what was going on, I set a border so I could
see the box.
body {
background: #cc9;
margin: 0px;
border: 1px solid black;
}
This bears further investigation, but it will have to wait until I
finish the switch.
Update
This turns out to be a difference in the way Firefox handles XHTML vs.
HTML. If I saved the document to a file with the .html extension, it looked
just fine, but if I named it .xhtml, I would see the weird, white border.
I moved the background-color directive out the the
body section and into the (newly created) html
section of the stylesheet, and it looks fine. The border disappears.
W3C may have something to say
about this issue, but it's hard to tell without a stardardese-to-english
translator.
