Mike's blog | Atom

Archive:
Tuesday, 08 September 2009

Porting to Loathsxome.
23:42

So Abhijit Menon-Sen has done some work on Blosxom, and the result is new software called Loathsxome, which I've decided to start using.

I like Blosxom's basic idea: it uses the filesystem rather than a database, and it's small and thus not awfully complicated. But the original author has abandoned it, and I wasn't too keen on the community-supported version (after finding a couple of bugs in it almost immediately after upgrading). ams overhauled the code, and I find the result very approachable. For example, the plugin interface is nicely-defined, though it will likely need some work: Loathsxome runs plugins in alphabetical order, and plugins work directly on the data that interests them, so I expect the resulting dependencies will be kind of a pain in the butt. On the other hand the plugins are generally simple enough that it's easy to see what order things need to run in. Some conventions for where they store data in Loathsxome's run-time state, and for declaring what they'll do, are probably 85% of what's needed.

As an aside, Rael's current blog is entirely Twitter-driven (apparently he is a user experience engineer at Twitter) and is a great example of how medium shapes message: With only 140 characters available, Twitter tends to vacuity and trivia. Or slower IRC.

Wednesday, 09 September 2009

Back to the mines.
01:33

So. A couple of weeks ago (before some travel) I went and bought some 13"x19" Ultra Premium Photo Paper Luster. Epson claims "Highest color gamut available for vivid color reproduction" - examining the profile shows it has the same gamut as the ultra premium special edition photo paper glossy (who invents these names?). In fact I kind of think these are the same profiles, just renamed.

In any case, it's reasonably close to the results of soft proofing. So I've been doing some work with it - first with my Russian Hill pic,

and now with a recent shot of the Golden Gate Bridge (which Kendra and I shot last night):

Bridge Tower

And they've come out fairly well - in particular the first one has given me a really lovely blue for the sky. They're not perfect, but good. A lot of the colors wind up out of gamut, though. There are a few ways to deal with that.

You can mess around with saturation, but that's going to suck, because you might have to desaturate a lot. Blah.

You can also lighten the image - this seems to bring the color back into gamut. It seems like it'd be useful to have some program which actually compares image and output gamuts and thus could suggest specific things to do to bring the colors all back in-gamut.

Futzing around with the hue could maybe work, but... yeah, dunno about that.

Note that select by color range offers "out of gamut colors", which I spent some time looking for (elsewhere of course), which is enormously convenient. If they're all mostly the same color, anyway. If that gives you two different colors then you probably don't want to process them the same way...

I've been sharpening with PS's Smart Sharpen filer, and printing test strips or sections with different amounts of sharpening. Note that it helps to make your test sheet image the same resolution as your actual image when doing that. I have to say that learning how to sharpen effectively sounds really tedious and uninteresting. I'm tempted to shell out for PhotoKit sharpener just to avoid learning the half-dozen sharpening options in PS and the near-infinitude of behaviors in each.

Anyway, more later (of course). But I'm back.

Saturday, 12 September 2009

Creativity and doing something new.
02:09

Jedrek and I are talking about creativity, and I'm wondering about people who do the same thing over and over. It seems quite boring, though I understand it's important in practice for artists if they want to sell. I remembered this quote from Edward Weston:

Wednesday the exhibit opened. I have had — as last year - applause; real homage - yes, even more than last year — and as before, the men far outnumber the women in attendance and interest. I am pleased, not satisfied, with my prints as they display themselves to me on the wall. No question but that I have gone ahead. And then comes the question, what next? An exhibit is always a climax to a certan period: Once shown, a print becomes definitively a part of one's past; if not actually discarded, it is relegated to a portfolio of old loves, to be referred to at times with perhaps no more than tender memories.

— The Daybooks of Edward Weston

NewsFire and why you shouldn't use it.
14:34

It looks like Loathsxome's RSS flavour has a bug - new posts cause Google Reader to show all the posts as new (is there some UUID stuff missing?). So I thought I'd try the blog in another feed reader. NewsFire was the first Google hit, and so I downloaded it and fired it up.

The first thing NewsFire did was nag me to upgrade to Snow Leopard (no thanks). It comes helpfully preloaded with a bunch of feeds - e.g. the BBC seemed to be in there. Not wanting any of those I deleted them all - except the newsfirex.com blog. Delete wouldn't work.

Nonplussed, I quit the app, edited ~/Library/Preferences/org.xlife.NewsFire.plist, and removed the entry from the Feeds array. But NewsFire helpfully readded it.

plonk Into the trash it goes. I'm won't use software which displays such arrogance. Nobody should.

Tuesday, 15 September 2009

RSS feed.
01:05

Each new blog post was causing all the previous posts to appear as new in Google Reader and in Vienna. Looking into it, I found that the pubDate elements were all just ":00", because loathsxome wasn't setting the variables the RSS template uses. So I added them to my customized fulldate plugin.

Notably, ams's blog doesn't have this problem. I assume he's using the default template, which doesn't include pubDate at all, and thus sidesteps the problem - so maybe I should just rip it out.

RSS feed, take two.
01:06

Yeah. That seemed to fix it.

RSS feed, take three.
01:37

Or... maybe not?

RSS feed, take four.
01:59

The real bug: escaping $url multiple times. The leading tilde in "/~mike/blog" was first escaped into %7E, and then into %257E, and then into %25257E... Etc. Adding a new post prepended another %25 to all the old posts' URLs, so the readers considered them new.

So I moved the entity encoding regular expressions and encoding of $url to before the foreach my $entry loop. That seemed to fix it... Except it didn't work in FireFox, because FireFox fetched the blog as "/%7Emike", which is correct according to the standard, but of course loathsxome escaped the leading percent, and everything was "/%257Emike/". So setting my $url explicitly was part two of that fix.