March 2015 - Talk - lists.gtalug.org

meeting tomorrow: topic?
by D. Hugh Redelmeier 15 Jul '15

15 Jul '15

According to <http://gtalug.org/> : "GTALUG is currently looking for speakers for the 9th December, 2014 meeting. If you are interested in speaking please contact our talks coordinator." Let's brainstorm on this. Here's a poorly organized dump of some of my thoughts. - I'd love to hear Lennart talk about what he's learned working on routers. - I'd like to learn what each of you has learned using Linux (or perhaps, not able to do using Linux) and what your current Linux-related interests are. - I'd like to hear how people have set up their internet gatways. What's worth doing, what's easy, what's hard. (It's time for me to revisit this after having a pretty static setup for a decade. I'd be willing to talk about what I've done but it is pretty stale.) - I'm interested in setting up my in-house services like a cloud. Perhaps <http://owncloud.org/>. Motivation: I want to keep control of my own data as much as possible. Can anyone speak to this? - Subproblem of above: I run MediaWiki already. Tips and tricks (eg. from Drew) about care, feeding, and use would be welcome. - I'd be willing to share 10 minutes worth of what I've learned about little Windows gadgets as Linux platforms. - I'd love to discuss whether this is interesting and useful <http://www.banana-pi.com/eacp_view.asp?id=64> This is a version of the Banana Pi with an added ethernet switch: 5 x 1G eithernet ports. The Banana Pi is intended to mimic the Raspberry Pi in some ways but with a stronger processor; my impression has been that this is goofy and that the Cubieboard2 and Cubietruck <http://cubieboard.org/> were more transparent uses of the Allwinner chips. But gluing on a switch seems very attractive. - chatting about peoples' workflow always seems useful: it's not just the tools, its how we use them, how we find them useful, and where they don't quite meet expectations. Examples: backup methodology, distro update methodology, favourite editor (leave out tribalism), programming language, etc. - why Go is interesting <https://golang.org/>. Why Rust is interesting <http://www.rust-lang.org/>. Why Python 2.x still has a loyal following. - what are you doing with your _____? Example: Raspberry Pi. - How you make your mobile device work with your desktop? Summary: we've had Question and Answer sessions, how about an Answer and Question session? By that I meen: people give a mini-chat about something they think is interesting and have the rest of us ask questions in response.

24 52

Dan Gilmore on computing freedom
by David Mason 16 Apr '15

16 Apr '15

https://medium.com/backchannel/why-i-m-saying-goodbye-to-apple-google-and-m…

5 6

[VOTE] Where should we have dinner before the April 2015 GTALUG meeting? (was Dinner Ideas for April's Meeting)
by Myles Braithwaite 07 Apr '15

07 Apr '15

<http://strawpoll.me/3951772>

2 4

git questions
by Giles Orr 07 Apr '15

07 Apr '15

I've started using git fairly heavily (although not necessarily very well). I have several repos: a couple are code, also my ~/.vim/ folder and I'm thinking about adding ~/bin/ . If I remember to push from a particular machine before leaving it, I'll never have to deal with merging as I'm the only user. While most likely my discomfort with merging will be overcome by practice as I understand git does it well (I use SVN at work: merging is shudder-inducing), I'm guessing keeping everything up to date is still preferable. These questions are mostly about incorporating information about the repo into the Bash prompt: I was impressed recently by the way zsh appears to handle it, with a sequence of tiny icons in the lower right corner of the terminal indicating relative status. I didn't talk to the zsh user long, so I don't know if that's built-in, a plugin, or something he did himself. I'm starting from some code I got from nitrous.io, lovely in its conciseness: parse_git_branch () { git branch --no-color 2> /dev/null | sed -e '/^[^*]/d' -e "s/* $.*$/\1$(parse_git_dirty)/" } parse_git_dirty () { git diff --no-ext-diff --quiet --exit-code &>/dev/null || echo "!" } Put \$(parse_git_branch) into your prompt and it tells you what branch you're on, and if there are any unstaged changes (and goes away if you're not in a repo). I'd prefer it did uncommitted rather than unstaged, haven't tried to fix that yet. But I want to update it as it doesn't deal with origin at all. So, the questions: What is the easiest and most concise way to determine if your local is behind origin master? I've found that "git remote show origin" will show this information, but I'm not sure if it's the "best" way to find out, and I'm also concerned that running that every time your prompt comes up would slow things down as it makes a remote call(?) to get an answer - when you might not even have a network connection, or worse, a very slow connection. "git status" usually says "Your branch is up-to-date with 'origin/master'" (or "ahead"), but occasionally - even though origin is configured properly - this line doesn't appear. Is there a way to convince it to always show this line? Unfortunately, "git status" doesn't seem to ever notice if you're "behind" origin, thus the need for "git remote show origin". Any fix for that? The thought was to have output for the prompt that looked like this: (everything synced): "master-" (uncommitted local changes): "master!-" (behind origin): "master^" (ahead of origin): "masterv" (ahead and behind, with local changes): "master!^v" You get the general idea. I'm aware git is capable of immense complexity (branches, detached head, multiple remotes, different remotes for push and pull, etc.) that aren't addressed here. If I tried to tackle all of that at once (especially given I don't understand most of it) my head would explode and nothing would get done. So I'm sticking with my simple use case until I have to deal with the more complex stuff. Any thoughts welcome. -- Giles http://www.gilesorr.com/ gilesorr(a)gmail.com

13 16

Youtube hints
by Walter Dnes 02 Apr '15

02 Apr '15

Given that Flash is being minimally maintained for linux, it's nice that Youtube has gon over to HTML5, but there are still a few hiccups that I ran into. I'll detail how I got around them, and also mention random access to videos. For the purposes of this example, I'll use the "Ancient Aliens Debunked" video, because it's over 3 hours long, which allows for examples of time offset. Warning... some URL-tweaking follows. Item 1) Youtube is now "officially" viewable in html5 without Flash, but I do run into a few instances of... ======================= X This plugin is disabled Manage plugins... ======================= Here's a video that "doesn't play on HTML5". https://www.youtube.com/watch?v=j9w-i5oZqaQ&hd=1 Let's break it up into sections. For URLs, the first parameter is preceeded by "?", and additional parameters are preceeded by "&". https://www.youtube.com/watch?v=j9w-i5oZqaQ&hd=1 breaks up into https:// www.youtube.com domain /watch "watch" parameter ?v=j9w-i5oZqaQ video ID &hd=1 HD mode selection If you get rid of the "&hd1", the URL becomes... https://www.youtube.com/watch?v=j9w-i5oZqaQ ...and it plays perfectly in HTML5, without Flash. Note that this is "the basic Youtube video URL". Item 2) OK, you've got a long video, and you want to jump a certain point in the video, without having to wait an hour plus to get to that point. This is done by appending a "&t=" parameter. It can contain "h", "m", and "s" subparameters. E.g. 1 hour, 20 minutes, and 15 seconds is signified by "&t=1h20m15s". You can include any or all of h/m/s. A few examples follow. "Pumapunku" (begins at 3 minutes 34 seconds into the video). https://www.youtube.com/watch?v=j9w-i5oZqaQ&t=3m34s "The Pyramids" (begins at 22 minutes 40 seconds into the video). https://www.youtube.com/watch?v=j9w-i5oZqaQ&t=22m40s "Easter Island" (begins at 1 hour 1 minute 35 seconds into the video). https://www.youtube.com/watch?v=j9w-i5oZqaQ&t=1h1m35s "Pacal's Rocket" (begins at 1 hour 5 minutes 43 seconds into the video). https://www.youtube.com/watch?v=j9w-i5oZqaQ&t=1h5m43s -- Walter Dnes <waltdnes(a)waltdnes.org>

4 3

Chromebooks Chromebits
by Daniel Wayne Armstrong 31 Mar '15

31 Mar '15

Upcoming cheapie ARM-powered Chrome devices looking to provide good homes for full-featured Linux installs: http://chrome.blogspot.ca/2015/03/more-chromebooks-for-everyone.html -- (o< .: Per curiositas ad astra .: http://www.circuidipity.com (/)_

4 3

war story: memtest86+ vs. UEFI
by D. Hugh Redelmeier 31 Mar '15

31 Mar '15

Summary: it seems that memtest86+ won't run under UEFI. You need to switch to "legacy" boot and use an appropriate CD/DVD/USB stick to supply memtest86+. I often run memtest86+ when I get new RAM. This increases my confidence that the RAM isn't busted and is compatible with the machine it is installed in. Running it overnight seems long enough without interfering with my use and enjoyment of the computer. I even run it on a brand-new machine as one of the confidence building measures. Well, that's what I used to do. I've not taken the time to do it recently. But I wanted to run it last night because I bought some (mislabelled!) RAM at the NCIX warehouse sale on Saturday. Ubuntu lets you install memtest86+ as a package (it may be a default package, I don't know). But you cannot run it under Ubuntu, you have to boot into it. When you install the package, it adds a grub entry to let you boot into memtest86+ instead of Ubuntu. Yet on my machine there was no such entry. Reading the script that generates the entry, I found that memtest86+ must be booted in 16-bit mode, i.e. old-fashioned BIOS, and not UEFI. So the script doesn't bother to generate the entry on a UEFI system (silently). To run memtest86+, you have to switch the system firmware to use "legacy" booting (old-fashioned BIOS/MBR). On my machine, you cannot enable both MBR and UEFI booting: you need to choose one. Once I switched to legacy booting, I could no longer use the Ubuntu on my hard drive since it was set up for UEFI. I used a Ubuntu installation DVD -- it has an early option for running memtest86+. It turns out that Fedora 21's live/install image has a newer memtest86+ than Ubuntu 14.04: 5.01 vs 4.20. The newer one lists the DIMMs or SODIMMs present on the system.

3 2

SSH key on a USB stick
by Giles Orr 30 Mar '15

30 Mar '15

I'd like to keep an SSH key on a USB stick rather than storing it locally on every machine I want to use the key on. The theory is that this is "more secure" for laptops that could be stolen or lost, or used by others (ie. at work). The problem is that when I run "ssh-add /media/stick/my_dsa" ssh refuses to use the key because the security permissions are too liberal. As you can probably guess, this is because the USB key is formatted as vfat and I'm trying to use the key on a Unix system - the problem was actually encountered on a Mac laptop, I'm assuming the behaviour would be the same with a Linux machine - even if it's not, I need a solution for the Mac. vfat doesn't support per-user permissions, and so the stick is mounted such that all files have 777 perms. I did some research which told me there's no way to tell ssh to ignore the perms error. I've seen a variety of solutions for this, but none are simple. I don't want to make a local copy of the key every time I have to add it. I don't want to have a separate ext4 partition on the key - not only is that awkward, but the ownership of the key is going to be a problem given that the UID of my user on the Mac won't be the same as it is on Linux machines (and may not even be the same on all my Linux machines, although I've tried to make that so). Has anyone found an elegant/simple solution for this? -- Giles http://www.gilesorr.com/ gilesorr(a)gmail.com

9 9

Intel - secure boot
by R. Russell Reiter 29 Mar '15

29 Mar '15

Here's a link to a talk on secure boot exploits etc. There's a bit on exploits which come in over the video aperture in order to flip the bits and get write permission on the secured stack, which I found interesting. https://m.youtube.com/watch?v=QDSlWa9xQuA -- Sent via K-9 Mail.

3 3

Help with if else statement
by Tim Carroll 29 Mar '15

29 Mar '15

Sorry if this type of post does not belong here. I'm a newbie to the group! I would like to write a script like the following but using while loop instead of for. I'm hoping that a while loop with an if else in it would rerun an fsck on only the drives that returned "***** FILE SYSTEM WAS MODIFIED *****" not rerun it on all of them! The if else bit has me stumped though... somewhere I need a "if [ $? -eq 0 ]; then" Any help would be greatly appreciated. #!/bin/bash # umount -a # df -hT echo -e '\n\tCAUTION!!! are all the disks unmounted!\n' read -p ' hit [enter] to continue or ^C to exit' read -p ' are you Sure? hit [enter] to continue' # get a list of disks attached ls -lh /dev/disk/by-label/ | sort -k 9 > disks-by-label-new.txt # clean it up with awk awk '$8 ~ /svr2arch/ { print $10; }' < disks-by-label-new.txt | awk -F '/' '{ print "/dev/"$3; }' > fsck-list-new.txt # this gives me a list like this "cat fsck-list-new.txt" /dev/sdb1 /dev/sdc1 /dev/sdd1 /dev/sde1 /dev/sde2 /dev/sde3 /dev/sde4 /dev/sdf1 /dev/sdf2 /dev/sdf3 /dev/sdf4 /dev/sdg10 /dev/sdg11 /dev/sdg12 /dev/sdg5 /dev/sdg6 # now start running the fsck on all disks for i in $(cat fsck-list-new.txt); do e2fsck -fvy -C 0 $i; done or while $(cat fsck-list-new.txt); do e2fsck -fvy -C 0 ; done -- _______________________________________________

4 4