Why did you remove count-lines-region and make the only option count-lines-page? How is counting Control-L-delimited “pages” useful as a default behavior?

Worse, the only way to get line-count from a region is now count-words-region, but that takes extra keystrokes to avoid invoking count-words. (yes, I know there are keyboard shortcuts; there are lots of keyboard shortcuts. I’ve never trusted them to be stable)

So, more workarounds in my .emacs file to turn Emacs back into a text editor. 😡 💩 🔥

My work MacBook Pro (mid-2014 edition) only has 512GB of storage. I had tried to order it fully loaded, but somebody upstream fainted from shock, despite the fact that my previous machine was nearly 8 years old.

I’ve got to do a lot of tinkering with VMware and Docker images for working out the IT infrastructure in our new building, and there’s just not enough space for it all on the internal drive. And since Apple solders everything to the motherboard these days, there’s no way to upgrade without replacing the machine. Browsing for small external SSDs, the Samsung T3 seems to be well-rated.

The only problem I can report so far is that it draws more power than the left USB3 port on my Mac can reliably supply. Moving it to the right port got rid of the eject/remount every 2-3 minutes. I’m not using their encryption, because I don’t need it; also, there are some reports of the supplied Mac software being flaky, and I had a few bad experiences with the way the equivalent WD driver works.

By the way, if you use Docker on a Mac, it will graduallyrapidly fill your disk unless you periodically rebuild the Docker.qcow2 file (works best if you supply the names of the images you don’t want deleted). A fix is reported to be in the works.

If you drag URLs around on a Mac, you get .webloc files. If you drag text around (successfully…), you get .textClipping files. Getting data back out of them can be annoying; some applications treat them as attachments, and most insist on grabbing the rich-text version of a clipping rather than the plain text.

webloc () {
  /usr/libexec/PlistBuddy -c 'print URL' "$1"
}
clipping () {
  DeRez -noResolve -only utf8 "$1" |
    perl -e 'while (<>) {
      chomp;
      next unless s/^\s+\$"([0-9A-F ]+)".*$/$1/;
      tr/ //d;
      $x.= $_;
    }
    print pack("H*",$x),"\n"'
}

…how much faster home Internet gets when you replace a flaky switch. It’s annoying that I had to drive to three different stores to find one (Office Max even had PoE-capable and 24-port switches, but I only needed a basic 8-port today).

I went from 1.4 Mbps down/0.88 up to 58/12; one of the nice things about having an OpenBSD router is that I can run iperf3 on it and verify that it’s not Comcast’s fault before opening a ticket.

If your Windows Update lasts for more than four hours, consult KB3161608.

Downside: while this resolves the dependency-calculation problem that caused update runtimes to go insane, it also manages to include changes that break a wide assortment of software. Because sure, why not force users to choose between business-critical applications and the ability to install security patches in less than a week of runtime?

No. No. I do not want a cloud-connected keyboard. I don’t care if the mechanical keyswitches are clad in the purest shimmering samite.

(via)

So I’ve got family in for the holidays, and my sister promptly took over the office for conference calls and video editing, leading to a distress call last night when the 2.2GB upload of the final cut of the new promo promised to take 70+ hours. (iperf to Hurricane Electric was running at about 500 Kilobits/sec upload speed, with ~10% packet loss)

I called Comcast at 6:30pm. They had a guy onsite at around 8:30pm, who verified that all my wiring was good, replaced the modem, and spent a merry twenty minutes poking through the flooded cable box out at the curb (he appreciated my big golf umbrella). All to say, “well, the problem isn’t here”, and he promised a network guy would take over in the morning.

Sure enough, by the time I got up and had breakfast, iperf was reporting 60 down, 12 up, which is a bit higher than before the outage.