Garmin Woes

I've had this problem a few times now: I'll record an activity on my 310XT, go to upload it later, and the sync agent fails to pull the activity.  My suspicion is that there is a bad block in the flash memory on the watch, and the watch hardware is old enough where it can't work around it.  I had this problem with the old Garmin ANT+ Agent under OS/X, and I just experienced it under Linux, with Garmin Forerunner 610 Extractor, a Python script that does a similar job to the ANT+ Agent.  What's interesting is that the new script always downloads all of the activity data, so I've been keeping the old activities around, and I now know that the error occurs somewhere between 540k-584k of memory usage, and that the value equates to 24% of the watch's memory, based on whah the watch says.  So it's a good guess that the watch has about 2 megs of memory on board.

(The sizes above are the byte count and the filesystem size of the downloaded activities.)

The solution here is to wipe out the old activities (and occasionally reset the watch), but it's still frustrating.  Luckily I only lost one run; there are times when I'll go for a few weeks without downloading any run data.

Appliance Frustrations

I have spent far too much time staring at this view, trying to figure out exactly what is wrong with our dishwasher. In a nutshell: the door switch is malfunctioning. It registers that the door is open when it isn't, sometimes in the middle of a cycle.  Last week I replaced the plunger, the spring, and the case cover, and tried to replace the relays, but ended up getting foiled by the wire connectors.

I finally got some proof tonight that the relays were a likely suspect: I was able to repeatedly trip the relays with the power on, and the machine never beeped. (In hindsight, I could have just tested them.)

I had been working on releasing the clasp portion of the relay connector, which is shown as a circle on each of the connectors above. What I actually needed to do was use a pair of needle nose pliers to loosen the teeth below, and then pop the clasp. I didn't think that they could exert that much tension on the terminal, but I was mistaken. I was about to just make up new connectors when I finally figured that out. That would have been interesting: I had 3 suitable connectors in my electrical repair toolbox, so the 4th would have ended up being a hack or an alligator clip until I could replace it with something better.

The machine is running now, as I write this. I really hope I've got it worked out, but I refuse to celebrate yet. This problem tends to resolve temporarily after opening the machine up. All of our major appliances have needed attention over the past month or so. The washing machine is still on the blink, though I might try to tackle that next. Our attempt at purchasing a replacement failed miserably, and the quote for repair was nearly half the cost of the machine itself.

In today's economical climate, with companies cutting back as much as they can, I offer this advice: buy the longest warranty that you can afford. Chances are good that it'll be cheaper than a single repair call.

Anyway, that's it for now. I've got a race report to write up, tentatively titled "Ouch", and I still need to finish the cleaning. Until next time, be excellent to each other.

Resetting Keyboard Shortcuts in GNOME 3

1. Run dconf-editor.  You can do this by opening a terminal and typing "dconf-editor" and hitting enter, or by searching for an app called "dconf Editor".
2. Navigate to: org -> gnome -> desktop -> wm -> keybindings.  You may see this written as "org.gnome.desktop.wm.keybindings", which means the same thing.
3. Scroll through the list on the right.  Any settings in bold have been changed.
4. For each changed setting, click first on the setting, so that it's highlighted, and then click "Set to Default".
5. Rinse, lather, repeat.
6. Once you are done with that, you should also check org.gnome.settings-daemon.plugins.power, and org.gnome.settings-daemon.plugins.media-keys.

Thanks to this help page for pointing me in the correct direction!

Geek: Dealing With .vnt Files

I switched this year from an iPhone to an Android phone, for a few reasons, which I might get into at some point in the future.  I use the Memo app on 'Droid to record notes; very often I use it to keep track of my weight before a hike, as well as the weight of my gear.  Getting data out of the app proved to be annoying.  I can export it to my computer through Dropbox, but it's in a .vnt format, that looks an awful lot like a vCard format, but I couldn't find any pre-built software capable of parsing it.  I was, however, able to come up with a quick workaround.

N.B.: this post assumes that you have a basic working knowledge of Linux, FreeBSD, or some other Unix or Unix-like system.  It assumes that you can install packages (vpim) on your system, and it also assumes that you know HOW to install packages (or compile software from distributed source code).  It also assumes that you have already have exported the Memo that you wrote on your Android phone into a .vnt format file.  It assumes that you are staring at that file, debating whether or not to just do a search and replace on those =0D=0A characters, and wondering why the Memo app couldn't just export a text file.

The .vnt file looks like this:

[doug@airship-pirate work]$ cat 2012-11-09.21.49.18.vnt
BEGIN:VNOTE
VERSION:1.1
BODY;CHARSET=UTF-8;ENCODING=QUOTED-PRINTABLE:Test=0D=0A=0D=0AMessage=0D=0AHere
DCREATED:20121109T214918
LAST-MODIFIED:20121109T214918
END:VNOTE
[doug@airship-pirate work]$ 

So, the content is in there, "Test Message Here", but it's a bit hard to work with.  You have a few options for translating this.  The most robust way is to use a vNote reader, but I have yet to find one, or have reason to write one.  So we'll use a vCard reader instead.  The package "vpim" provides a few utilities, including vcf-dump, which we'll exploit for our own nefarious purposes in just a moment.  First, though, we have to tweak that VNOTE into a VCARD.  If you don't do this step, vcf-dump gets all cranky.

[doug@airship-pirate work]$ sed -e 's/VNOTE/VCARD/' 2012-11-09.21.49.18.vnt > 2012-11-09.21.49.18.vcf
[doug@airship-pirate work]$ head -1 2012-11-09.21.49.18.vcf
BEGIN:VCARD
[doug@airship-pirate work]$ 

Easy enough.

Next, run it through vcf-dump:

[doug@airship-pirate work]$ vcf-dump 2012-11-09.21.49.18.vcf
..Begin="VCARD"
..Version="1.1"
..Body="Test\r\n\r\nMessage\r\nHere"
 ENCODING=[QUOTED-PRINTABLE]
 CHARSET=[UTF-8]
..Dcreated="20121109T214918"
..Last-modified="20121109T214918"
..End="VCARD"

Great.  vcf-dump handled translating that quoted-printable string, and presumably would handle other encodings as well (i.e. base64), without too much effort on our part.

Finally, the last step.  I'm going to grab that "Body" line, cut out the portion between the quotes, and then translate those CRLFs ("\r\n") into actual newlines. 

[doug@airship-pirate work]$ vcf-dump 2012-11-09.21.49.18.vcf | fgrep "Body=" | cut -f 2 -d '"' | sed -e 's/\\r\\n/\n/g'
Test

Message
Here

Voilà.

I hope that helps.  Please remember to back up any files prior to attempting to translate them, since one botched '>' can ruin your day.