Shallow Thoughts : : Apr

The evil HTML double-dash problem in Emacs is still there

Long ago (in 2006!), I blogged on an annoying misfeature of Emacs when editing HTML files: you can't type double dashes. Emacs sees them as an SGML comment and insists on indenting all subsequent lines in strange ways.

I wrote about finding a fix for the problem, involving commenting out four lines in sgml-mode.el. That file had a comment at the very beginning suggesting that they know about the problem and had guarded against it, but obviously it didn't work and the variable that was supposed to control the behavior had been overridden by other hardwired behaviors.

That fix has worked well for eight years. But just lately, I've been getting a lot of annoying warnings when I edit HTML files: "Error: autoloading failed to define function sgml_lexical_context". Apparently the ancient copy of sgml-mode.el that I'd been using all these years was no longer compatible with ... something else somewhere inside emacs. I needed to update it.

Maybe, some time during the intervening 8 years, they'd actually fixed the problem? I was hopeful. I moved my old patched sgml-mode.el aside and edited some files. But the first time I tried typing a double dashes -- like this, with text inside that's long enough to wrap to a new line -- I saw that the problem wasn't fixed at all.

I got a copy of the latest sgml-mode.el -- on Debian, that meant:

apt-get install emacs23-el
cp /usr/share/emacs/23.4/lisp/textmodes/sgml-mode.el.gz ~/.emacs-lisp
gunzip ~/.emacs-lisp/sgml-mode.el.gz

Then I edited the file and started searching for strings like font-lock and comment.

Unfortunately, the solution I documented in my old blog post is no longer helpful. The code has changed too much, and now there are many, many different places where automatic comment handling happens. I had to comment out each of them bit by bit before I finally found the section that's now causing the problem. Commenting out these lines fixed it:

   (set (make-local-variable 'indent-line-function) 'sgml-indent-line)
   (set (make-local-variable 'comment-start) "")
   (set (make-local-variable 'comment-indent-function) 'sgml-comment-indent)
   (set (make-local-variable 'comment-line-break-function)
        'sgml-comment-indent-new-line)

I didn't have to remove any .elc files, like I did in 2006; just putting the sgml-mode.el file in my Emacs load-path was enough. I keep all my customized Emacs code in a directory called .emacs-lisp, and in my .emacs I make sure it's in my path:

(setq load-path (cons "~/.emacs-lisp/" load-path))

And now I can type double dashes again. Whew!

Tags: emacs, editors, html
[ 12:42 Apr 29, 2014    More linux/editors | permalink to this entry | ]

Some code from PiDoorbell

If anyone has been waiting for the code repository for PiDoorbell, the Raspberry Pi project we presented at PyCon a couple of weeks ago, at least part of it (the parts I wrote) is also available in my GitHub scripts repo, in the rpi subdirectory. It's licensed as GPLv2-or-later.

That includes the code that drives the HC-SR04 sonar rangefinder, and the script that takes photos and handles figuring out whether you have a USB camera or a Pi Camera module.

It doesn't include the Dropbox or Twilio code. For that I'm afraid you'll have to wait for the official PiDoorbell repo. I'm not clear what the holdup is on getting the repo opened up.

The camera script, piphoto.py, has changed quite a bit in the couple of weeks since PyCon. I've been working on a similar project that doesn't use the rangefinder, and relies only on the camera to detect motion, by measuring changes between the previous photo and the current one. I'm building a wildlife camera, and the rangefinder trick doesn't work well if there's a bird feeder already occupying the target range.

Of course, using motion detection means I'll get a lot of spurious photos of shadows, tree limbs bending in the wind and so forth. It'll be an interesting challenge seeing if I can make the code smart enough to handle that. Of course, I'll write about the project in much more detail once I have it working.

It looks like the biggest issue will be finding a decent camera I can control from a Pi. The Pi Camera module looked so appealing -- and it comes in a night version, with the IR filter removed, perfect for those coyote, rabbit and deer pictures! -- but sadly, it looks like its quality is so poor that it really isn't useful for much of anything. It's great for detecting what types of animals visit you (especially at night), but, sadly, no good for taking photos you'd actually want to display.

If anyone knows of a good camera that can be driven from Linux over USB -- perhaps a normal digital camera that supports the USB camera protocol? -- please let me know! My web searches so far haven't been very illuminating.

Meanwhile, I hope someone finds the rangefinder and camera driving software useful. And stay tuned for more detailed articles about my wildlife camera project!

Tags: raspberry pi, speaking, conferences, maker
[ 11:57 Apr 23, 2014    More hardware | permalink to this entry | ]

Back from PyCon

I'm back from Montreal, settling back in.

The PiDoorbell tutorial went well, in the end. Of course just about everything that could go wrong, did. The hard-wired ethernet connection we'd been promised didn't materialize, and there was no way to get the Raspberry Pis onto the conference wi-fi because it used browser authentication (it still baffles me why anyone still uses that! Browser authentication made sense in 2007 when lots of people only had 801.11g and couldn't do WPA; it makes absolutely zero sense now).

Anyway, lacking a sensible way to get everyone's Pis on the net, Deepa stepped as network engineer for the tutorial and hooked up the router she had brought to her laptop's wi-fi connection so the Pis could route through that.

Then we found we had too few SD cards. We didn't realize why until afterward: when we compared the attendee count to the sign-up list we'd gotten, we had quite a few more attendees than we'd planned for. We had a few extra SD cards, but not enough, so I and a couple of the other instructors/TAs had to loan out SD cards we'd brought for our own Pis. ("Now edit /etc/network/interfaces ... okay, pretend you didn't see that, that's the password for my home router, now delete that and change it to ...")

Then some of the SD cards turned out not to have been updated with the latest packages, Mac users couldn't find the drivers to run the serial cable, Windows users (or was it Macs?) had trouble setting static ethernet addresses so they could ssh to the Pi, all the problems we'd expected and a few we hadn't.

But despite all the problems, the TAs: Deepa (who was more like a co-presenter than a TA), Serpil, Lyz and Stuart, plus Rupa and I, were able to get everyone working. All the attendees got their LEDs blinking, their sonar rangefinders rangefinding, and the PiDoorbell script running. Many people brought cameras and got their Pis snapping pictures when the sensor registered someone in front of it. Time restrictions and network problems meant that most people didn't get the Dropbox and Twilio registration finished to get notifications sent to their phones, but that's okay -- we knew that was a long shot, and everybody got far enough that they can add the network notifications later if they want.

And the most important thing is that everybody looked like they were having a good time. We haven't seen the reviews (I'm not sure if PyCon shares reviews with the tutorial instructors; I hope so, but a lot of conferences don't) but I hope everybody had fun and felt like they got something out of it.

The rest of PyCon was excellent, too. I went to some great talks, got lots of ideas for new projects and packages I want to try, had fun meeting new people, and got to see a little of Montreal. And ate a lot of good food.

Now I'm back in the land of enchantment, with its crazy weather -- we've gone from snow to sun to cold breezes to HOT to threatening thunderstorm in the couple of days I've been back. Never a dull moment! I confess I'm missing those chocolate croissants for breakfast just a little bit. We still don't have internet: it's nearly 9 weeks since Comcast's first visit, and their latest prediction (which changes every time I talk to them) is a week from today.

But it's warm and sunny this morning, there's a white-crowned sparrow singing outside the window, and I've just seen our first hummingbird (a male -- I think it's a broad-tailed, but it'll take a while to be confident of IDs on all these new-to-me birds). PyCon was fun -- but it's nice to be home.

Tags: conferences, speaking, python
[ 10:20 Apr 17, 2014    More conferences | permalink to this entry | ]

Snow-Hail while preparing for Montreal

Things have been hectic in the last few days before I leave for Montreal with last-minute preparation for our PyCon tutorial, Build your own PiDoorbell - Learn Home Automation with Python next Wednesday.

But New Mexico came through on my next-to-last full day with some pretty interesting weather. A windstorm in the afternoon gave way to thunder (but almost no lightning -- I saw maybe one indistinct flash) which gave way to a strange fluffy hail that got gradually bigger until it eventually grew to pea-sized snowballs, big enough and snow enough to capture well in photographs as they came down on the junipers and in the garden.

Then after about twenty minutes the storm stopped the sun came out. And now I'm back to tweaking tutorial slides and thinking about packing while watching the sunset light on the Rio Grande gorge.

But tomorrow I leave it behind and fly to Montreal. See you at PyCon!

Tags: raspberry pi, python, hardware, programming, speaking, conferences, maker
[ 18:55 Apr 06, 2014    More misc | permalink to this entry | ]