I messed up my automation of backups, meaning that after two years my entire VPS had secretly filled up. This lead to MariaDB being unable to initialize. After taking care of the root cause MariaDB still refused to start.
$ sudo tail -3 /var/log/mysql/error.log
2018-02-20 12:07:45 140649776292416 [Note] Recovering after a crash using tc.log
2018-02-20 12:07:45 140649776292416 [ERROR] Can't init tc log
2018-02-20 12:07:45 140649776292416 [ERROR] Aborting
Stupidly, just removing the zero byte `/var/lib/mysql/tc.log` file took care of the problem.
Gource videos have been popping up on my radar for several years now, but for some reason I decided to play around with it yesterday. My tweaks were very minor: much increased speed, avatars, a logo, a background image, and some music to go along with it.
I started with the avatars. To grab them, I used this script with a quick replacement for the error handling:
if($rc != 200) {
unlink($author_image_file);
my $gh_url = "https://avatars.githubusercontent.com/".$author."?size=".$size;
warn "fetching gh image for '$author' $email ($gh_url)...\n";
my $rc = getstore($gh_url, $author_image_file);
next;
}
sleep(1);
That way users without Gravatars but a GitHub username in their commit message still get the one you’re used to seeing. I manually deleted all of the default GitHub no avatar images. That sounds like a lot of work, but you can just sort by size.
I did a quick inversion of the KOReader logo in GIMP. Nothing special there.
The background image I grabbed from Pixabay, a useful site for free photographs. The music I got from dig.ccmixter.org, something similar for music.
Note that I lowered the volume of the music, more on that here. You could also play around with fade-outs, whether in Audacity or straight from ffmpeg. Also see my post on lossless audio cutting for when everything is good but you just want to trim the audio a little. That’s all!
This is a little experiment with Gource. Direct link to video here. KOReader is a document viewer for E Ink devices. It supports PDF, DjVu, XPS, CBT, CBZ, FB2, PDB, TXT, HTML, RTF, CHM, EPUB, DOC, MOBI, and ZIP files. It currently runs on Kindle, Kobo, PocketBook, Ubuntu Touch and Android devices. More information here.
As one of the maintainers of KOReader, a versatile a document and image viewer for E Ink devices, I’m proud to announce the latest release.
Special thanks to @dengste this month for adding natural light support on the Kobo Aura One (#3679).
You can control the natural light feature on the Kobo Aura One through the regular frontlight widget and the frontlight gesture controller. The latter allows for controlling warmth by swiping on the right of the screen, and is accessible via Tools → More plugins → Frontlight gesture controller.
Those who read two-column papers might like the new two-column navigation (@FranMarelli This does not work in conjunction with scroll mode.
Those willing to experiment can try the new Open with… option (#3651).
Finally, a little clarification on last month’s HTML dictionary support. We use MuPDF to render the HTML dictionary results. Unfortunately, MuPDF expects its input to be well-formed XHTML, meaning XML. When the HTML is tag soup instead, KOReader will fall back to a stripped version of the HTML dictionary entry. Should this be the case, you can manually fix up the output by adding a .lua file in the dictionary directory. Further details and a few examples are provided by @poire-z in #3611.
We’d like to thank all contributors for their efforts. Some highlights include:
Half a decade ago I neglected to jot this down, so I had to figure it out again. Just in case someone landed here searching for the most basic R problem, you start the program using uppercase R, not lowercase r. Anyway, to work with a package in a git repository or some such the easiest method is the devtools package.
Without any parameters, the load_all() function from devtools loads the current directory without installing. You could effect the same with more keystrokes using load_all('./'), and of course you can pass any path instead of relying on the current working directory. But in combination with git I find it easiest to just stick with that. After you’ve made some changes in the source, just run load_all() again.
Eenoog sloop voorzichtig door het huis.
Waar was de vijand?
Waar was de verschrikking?
Waar was het kwaad?
Eenoog werd bekropen
door een naarstig gevoel.
Het kwaad miste ook een oog.