There are few names that drive as much web traffic in the western world right now than "Elon Musk" and "Tesla."

.

damn you prolog, I mean damn you university

Everybody should learn prolog. It's such a wonderful piece of idea.

If you express a graph as:

transition(S, a, T)
transition(S, b, U)
transition(T, 0, V)
transition(U, 0, K)
transition(T, c, K)
transition(U, a, U)
...

mark a special node as an accepting state

accept(c, K)

then validating input is only:

validate(S, [I|IS]) :-
  accept(I,S).
validate(S, [I|IS]) :-
  transition(S,I,X),
  validate(X,IS).

how daming is that.

Programmation par l'exemple en Caml par Louis Gacôgne - 25 euros merveilleusements depenses

Pour ceux qui ont deja des bases en programmation, et en programmation fonctionnelle, ce petit manuel est une mine d'exercises aussi concis que generaux.

arbres, reines, graphes, hasard, lambda calcul, derivation symbolique, jamais trop de details, jamais pompeux ni trop succint... genial

http://www.decitre.fr/livres/programmation-par-l-exemple-en-caml-9782729819101.html

local ssh neighbors ?

A little bash function leveraging nmap .. mostly naive and fragile.
Until I learn how to do tcp|ip discovery.

neigh_ssh ()
{ 
    sudo nmap -v -p 22 --open --host-timeout 1 192.168.1.0/25 \
  | grep --color=auto -E -o "open port 22.*[0-9]{1,3}\.[0-9]{1,3}\.[0-9]{1,3}\.[0-9]{1,3}" \
  | grep --color=auto -E -o "[0-9]{1,3}\.[0-9]{1,3}\.[0-9]{1,3}\.[0-9]{1,3}" \
  | sort \
  | uniq
}

systemd-nspawn and noexec flag

Trying to create customized archlinux iso (archiso, releng) I hit a lot of issues, most of them 'execv? failed, permission denied'.

Later on, just trying to run a systemd-nspawn container, again, execv permission. Boot fails.

Googling didn't get me very far, except to broaden the scope from systemd-nspawn to something more general. I try to run a binary (ls) from the container subtree. Permission denied. I md5sum container/.../ls and get the same md5 as the main linux install I'm running. I thought it could be the wrong architecture or a different binary. So if two binaries refuse to run .. maybe they're treated differently from the host OS. I then realize that my container and custom archiso were on a mounted drive.

mount | grep <mountpoint> -&gt; mount .... (...,noexec, ...)

EUREKA.

udevil mounts this drive with a lot of protective flags. udevil unmount <mountpoint>; sudo mount <device> /mnt; <container>/ ... /ls works.

systemd-nspawn -b -D <container> works.

Thanks to #archlinux for nothing (beside rubber ducking ;)</container></container></device></mountpoint></mountpoint>

mtp under [arch]linux: mtpfs vs gvfs-mtp [mount update]

Wanted to copy some file from a Moto G to my laptop. Installed mtpfs. Mounting took minutes and failed. As a last hope I tried gvfs-mtp. I don't know how to use the gvfs-* CLI though. But it's supposed to work out of the box in GTK apps. So Chromium + Ctrl-O. I can see the smartphone listed. Walk through the phone FS. It's instantaneous, snappy as I like it. But what about the CLI ?

Well gvfs-info says nothing. There's gvfs-mount, which says nothing. gvfs-mount --list outputs the mounted drives on this machine. And the last one is mtp://[usb:005,009]/ ... Hmm maybe gvfs-ls takes an URI ?

$ gvfs-ls mtp:/// # Error: The specified location is not mounted
$ gvfs-ls mtp://  # Error: The specified location is not mounted
$ gvfs-ls mtp:///sdcard # same as above

could the [usb:...] be mandatory ? odd syntax but alas

$ gvfs-ls mtp://[usb:005,009]/
Internal Storage # yay

$ gvfs-tree mtp://[usb:005,009]/ | head -n 32
mtp://[usb:005,009]/
`-- Internal Storage
    |-- .CM_Cloud
    |   |-- cm
    |   |   |-- cm_cloud_sdk_expand.db
    |   |   `-- cm_cloud_sdk_expand.db-journal
    |   `-- cmqp
    |       |-- cm_cloud_sdk_expand.db
    |       `-- cm_cloud_sdk_expand.db-journal
    |-- .DataStorage
    |   `-- ContextData.xml
    |-- .UTSystemConfig
    |   `-- Global
    |       `-- Alvin2.xml
    |-- .com.zwsoft.zwcad
    |   |-- local_Thumb
    |   |   |-- Fonts
    |   |   `-- Sample Drawings
    |   |       |-- Villa.dwg.png
    |   |       |-- ar_house_2.dwg.png
    |   |       |-- crane.dwg.png
    |   |       `-- l_1a.dwg.png
    |   |-- resources
    |   |   |-- adinit.dat
    |   |   |-- arialuni.ttf
    |   |   |-- plotstyle
    |   |   |   |-- Icad.ctb
    |   |   |   |-- Monochrome.ctb
    |   |   |   |-- zwcad-Color.stb
    |   |   |   |-- zwcad.ctb
    |   |   |   `-- zwcad.stb
    |   |   |-- zwcad.dwt

Didn't try to copy anything yet, I guess it won't be worse than doing a full http roundtrip over wifi as I did before.

ps: there was a side effect of trying gvfs through GTK file open dialog. It mounted the volume on click. Thus the ability to use gvfs-* afterwards.

So there's a few things to do in cli beforehand.

source: http://stackoverflow.com/questions/483460/how-to-mount-from-command-line-like-the-nautilus-does

$ gvfs-mount --list
Volume(0): XT1032
  Type: GProxyVolume (GProxyVolumeMonitorMTP)

This shows XT1032 (Moto G) is detected but not mounted. What we want is the following:

$ gvfs-mount --list
Volume(0): XT1032
  Type: GProxyVolume (GProxyVolumeMonitorMTP)
  Mount(0): XT1032 -&gt; mtp://[usb:005,013]/
    Type: GProxyShadowMount (GProxyVolumeMonitorMTP)
Mount(1): mtp -&gt; mtp://[usb:005,013]/
  Type: GDaemonMount

Usually to mount a 'storage' device, one uses /dev/sd*. But there's no sd* node for the XT1032.
gvfs-mount --list --info lists a bit more info (sic) (errata: --list -i)

$ Volume(0): XT1032
  Type: GProxyVolume (GProxyVolumeMonitorMTP)
  ids:
   unix-device: '/dev/bus/usb/005/013'
  activation_root=mtp://[usb:005,013]/
  themed icons:  [phone]
  symbolic themed icons:  [phone-symbolic]  [phone]
  can_mount=1
  can_eject=0
  should_automount=1

And trying with the /dev/bus/... path

$ gvfs-mount -d /dev/bus/usb/005/013
Mounted /dev/bus/usb/005/013 at /run/user/1000/gvfs/mtp:host=%5Busb%3A005%2C013%5D

$ gvfs-ls mtp://[usb:005,013]/                                                                                                                  
Internal Storage

# yay

usb key replicate

* usb key duplicate

* partition table

  manual cgdisk

* rsync partitions

  rsync -auv --progress <source></source> <target>

* rsync caveat : hard links, sparse file

  warning: some sparse files may have huge virtual size
    rsync, will attempt to expand them fully
    i.e: docker devicemapper (60MB on disk, virtually 100G)

  warning: some programs have one fat binary with hard links
    rsync, will attempt to copy them fully
    i.e: git, which has 114 hard links with different names

* BIOS boot (ef02) partition

  dd if=/dev/sdX1 of=/dev/sdY1 bs=1M # simply

* grub

  sudo grub-install --target=i386-pc --debug --boot-directory=/<root-mountpoint>/boot/ /dev/sdY

  warning: do not mess the device names and mountpoint names as it may modify your host system grub config


* grub UUID

  grub menuentry refers to source key UUID

* adjust root UUIDs
  /etc/fstab still refers to source key UUID

  blkid /dev/sdX? &gt;&gt; /etc/fstab
  vi /etc/fstab

  (some sed-fu would be nice)</root-mountpoint></target>

nixos 16.03 live iso boot "directly" from grub

I hate extracting iso images. I prefer to run them through grub loopback. Nixos 16.03 release is a few days old, I couldn't resist. Alas, booting it this way drops me in a rescue shell.

The initrd is set up to only have one layer of virtual fs. So it assumes the squashfs nix-store would be in /mnt-root/iso and complains since it is not.

So after trying to MITM myself and mount the desired FS where they should be, I replicate the final steps of the nixos initrd init script. Kernel panic.

I'm dumbfounded, I need to be more thorough. The init scripts reads a fsinfo file (fstab-like in a different format say) to prepare the new root for the real stage 2 kernel.

btw, my mistake was to simply mount the nix-store squashfs file in /mnt-root/nix/.ro-store and then symlink /mnt-root/nix/store to it. It works only partially as the stage2 kernel find its init and proceed to run systemd but hangs after a few services doing nothing. Only option: CTRL-ALT-DEL. (I even tried to instrument stage2 systemd into booting the rescue target, hoping for a rescue shell to investigate the hang .. no luck). Then I realize that the fsinfo mounts the squashfs in nix/.ro-store BUT, also prepares a unionfs in nix/store ! Maybe stage2 is waiting for a unionfs mount in nix/store and my symlink is confusing things up.

So I insert the missing layer in there. And just sh /init. It fails horribly saying 'cannot stop udev/queue' then 'Killed'. Not thrilled.

I extract the final steps of the init script (populate /mnt-root with proc,sys,run,dev; kill a few things; restore the path for `modprobe` in the new /proc/sys/kernel/modprobe value). And then exec switch_root ...

I can now enjoy nixos 16.03 live booting directly from grub. If you allow directly to allow user intervention .. heh.

archlinux live archiso expand cow persistent overlay

Seems that by default, archiso cow persistent space is set to 262M. Too short for some needs [1]. Linux said the cow fs was filled while installing some packages. Struggle to find a way to add some more writeable space. Then I saw in mtab fs options that cowspace being tmpfs, has a size parameter. So remounting it with different settings.

sudo mount -oremount,size=<space>k cowspace

did the job

now there's an overlay named airootfs, it will show the old size, remounting put it in sync.

sudo mount -oremount airootfs

now both will show <space>.

quick note: php v7 phar and openssl modules are now in core

If you see warnings such as below, for openssl.so or phar.so:

PHP Warning:  PHP Startup: Unable to load dynamic library '/usr/lib/php/modules/openssl.so' - /usr/lib/php/modules/openssl.so: cannot open shared object file: No such fi
le or directory in Unknown on line 0
PHP Warning:  PHP Startup: Unable to load dynamic library '/usr/lib/php/modules/phar.so' - /usr/lib/php/modules/phar.so: cannot open shared object file: No such file or
directory in Unknown on line 0

Then comment out the entries in your php.ini

source: http://unix.stackexchange.com/questions/256338/how-to-install-php7-modules-in-archlinux/256482

ps: Similar issue with mysql.so, deprecated for mysqli.so. source: https://www.digitalocean.com/company/blog/getting-ready-for-php-7/

java8 nashorn javascript and clojure run into a repl

Java javascript engine (nashorn) is easily accessible through the javax.script fAPI. So I popped a clojure repl to instanciate a javascript compiler so Java could run my js bytecode.

(ns three.lisp.tower)

(def js (.getEngineByName (javax.script.ScriptEngineManager.) "nashorn"))

(defn jsc [code]
          (let [prelude "{add: function(a,b){return a+b;}, inc: function(n){return n+1;}}",
             link "(function(prelude){return (function(prelude){ %s })(prelude);})(%s)"
             emit (fn [l p s] (format l s p))
             js (fn [s] (.eval js s))]
        (js (emit link prelude code))))

user=> (jsc "return Array(1,2,3).map(prelude.inc).reduce(prelude.add)")

9.0

#EnoughProcrastination

frictionless Java with .. clojure

Late realization: you can use clojure with Java. Not with Java. For Java. We all know clojure java interop is good, but that's not what is at play(sic). See, Java is verbose (duh), compiled (duhbl) and mute. On the other hand, clojure, being a lisp with a repl, is real-time, short and generic, and doesn't even require `print` to inspect values most of the time. So instead of writing long main println-full methods, compiling, running, parsing the output ... you just import some java classes and start poking around. Frustration: gone. Even better, clojure is human friendly, has clojure.reflect/reflect to inspect classes and/or instances; clojure.pprint/pprint for quick formatting needs. If java bytecode came with docstrings and javadoc... clojure repl would be a better IDE than Eclipse. And it's probably a few lines away to scrape some remote or local html set and get going. People probably delegate that to manual browsing or a text editor extension. Anyway, my java project makefiles [1] have now a `repl:` rule to enjoy Java without Java (sorry dude).

[1] yes, I like IDE-free projects so a makefile with a few rules. I even learned about jdb, because yeah, text mode java debugging !

programming theorization and design based on information content ?

instead of copying and mutating data, layer and intercept ? lazy and virtual

adding two strings yields an entity knowing about the two strings and how to be a string based on the idea of concatenation, no need to recreate a new string of both content

str s \
         ,`- str u
str t /

it's all good in systems were information chunks aren't mutated, erased by local agents

user extension quality

I used to refactor (popular) wordpress plugins. I don't do it but I use a chromium extension (not a popular one though) which I realized was using half of my ram. Pretty hot for a leak.

The source is tiny, so tiny you wonder how this thing can leak at all, there's nothing except one simple loop.

Turns out it was a piece of debugging code. The whole production source is full of them, 90% of them are harmless except that one line.

User extensions...

System performance reminders - Overheating, slow disk IO, useless network IOs

In the last few months my laptop started to act differently, annoyingly slow, annoyingly hot. I didn't sense changes, reasons for that drop in 'performances'. I was also a bit exhausted so my first reaction was: this one isn't enough, buy a new one. In the back of my mind I knew that didn't make any sense, but still.

Bit by bit issues were resolved. Here's the log:

Browser ad block

I used the usual adblock plus chrome extension. It's subpar and sucked lots of cpu cycle and memory. Recently ublock Origin appeared and requires a lot less resources. Page load 2x faster, memory usage down.

SSD hysteresis

I have an intel 320series 80GB. Configured with auto-trim (fstab under linux). Never cared about trim anymore but when I started to reach less than 1GB free space, speed dropped. Welcome back to the days of mechanical HDDs. Articles say SSD needs free space to allow for wear leveling and such. So i tried to delete the biggest most useless files. Now above 4GB free, still no normal performance. trimming is useless since it's automatic, and sudo fstrim --all --verbose reports '0 bytes trimmed'.

I thought it was a linux bug, removed auto-trim and rebooted. Still nothing. But then manually trimming yeld some output 'xxGB trimmed'. Performance is back to normal levels. Still don't know if it was a bug, where, or if it's expected behavior that  when reaching <1GB free space, you have to force trim ...?

DNS

Installed pdnsd, a DNS cache. For heavy ajax websites like twitter, page rendering was significantly faster. Using twitter made me anxious(see the last point to see why), now I'm mostly zen. (2x, 3x faster perceived speed)

Heatsink / fan

The most recent one. The machine felt hot. But when the system shutdowns for no reason you start to worry. Logs have messages with cpu issued critical warnings about reaching maximum operating temperature -> shutdown. It was often around 80degC, which I thought was alright for CPUs but anyway. Peeking under the keyboard I see no large amount of dust. After a few weeks and 3 more panic shutdowns I tear down the whole thing. Still no big blog of dust that could impede air flow. I use the house vacuum cleaner, fold bits of paper to rub through heatsink fins. Now the cpu tops at 60degC when watching 720p. and below 50 on normal use. What a relief, and a simple fix too, too easy to forget.

I'm tempted to redo it but deeper. I can renew the thermal paste between cpu and sink. And also make new 'fin cleaners' with alcohol to clean the metal surface better.
People told me even a thin layer of dust on metal can change the thermal diffusion power.

On my way I also dismantled the keyboard entirely, it was disgusting. Alcohol bathed keycaps, brushed the board. The laptop feels almost new.

The joy of 'knowing' (a little, I'm still dreaming of fixing mainboards, fans DC motor, firmware and LCD panel)