The first thing to do with a new clean (empty) $HOME, is to populate it with my actual files. Not the configuration cruft that is of concern, but my actual “stuff”.

For my own use, these consist of a variety of directories that are a mix of . . . → Read More: new $HOME day 1: restoring my files]]> rebuilding $HOME: Days 1 and 2

The first thing to do with a new clean (empty) $HOME, is to populate it with my actual files. Not the configuration cruft that is of concern, but my actual “stuff”.

For my own use, these consist of a variety of directories that are a mix of traditional unix system style directory names (eg: etc, lib, var, bin, public_html, Maildir, sin (for tempting tmp files) – with expected purposes, and more modern GUI styled names (eg: Photos, Videos, Public, Desktop), as well as a few of my own naming (verb, ketchup (for sauce src files )

Restoring these has largely been a matter of a paranoid comparison between my old recovered $HOME and the last nightly backup (I now have many ~/*/.md5sums files!)

In many cases both versions of a directory were the same, so restore was easy. In a few cases, they were different and a series of `diff -r` between the backup and recovered directories clarified that the changes were only new files, not the corruption of old ones.

Without too much difficulty, some 218 of the 221gig of my original $HOME was restored.

Regarding dotfiles… I moved a few over quickly – .ssh .cliverc and .tmux.conf for example (the latter two moving into ~/etc with symlinks), and a few others were created in opening usages (.selected_editor, .lesshst, .cache

Once I start using a GUI again, the dotfile population will explode. I hope I have a greater measure of sanity before then.

Coming soon: shell and email configurations…

Specifically, jpegtran can (with . . . → Read More: Multi quality JPEG hack]]> tl;dr: In which I attempt to montage jpeg which comprises areas of differing internal jpeg quality, for fun and, er… artistic geeky interest.
You see, once upon a time, I was thinking about lossless jpeg editing. The sort that jpegtran can do.

jpegtran (1)         - lossless transformation of JPEG files

Specifically, jpegtran can (with some understandable limits) rotate/flip and mirror losslessly, and also crop – for values of “lossless” that only apply to the cropped image data.

I soon got to wondering if the reverse could also be true. If jpeg image data can be cropped losslessly; then it must not rely on surrounding data and so surely paste is equally possible.

I researched some and found… http://www.imagemagick.org/Usage/formats/#jpg_lossless. If you scroll down to the ‘Mixed JPEG Quality’ section it’s suggested:

* use jpegtran to merge the q60 on top of the q100

“So”, I thinks to myself, “…jpegtran can do that, huh?”

Turns out that it can’t. Not the version currently in Debian anyway.

…But an experimental version can. See the version that the chaps have here: http://jpegclub.org/jpegtran/

Only the original 1600x1024 is a true quality hybrid

The information on that page indicates to me that the ‘drop’ option has been experimental since 2000 at least… and I wonder why it’s not mainstream then. But, let’s take a look anyway. There are helpful binaries and a sample script provided, but I had something a bit more fun in mind.

One download later, and some fiddling around… and here is this! IMAG KWALITEE for all!

Counting from the bottom where it’s most evident, the rows of 80 pixels high each are of quality 1, 4, 5, 6, 7, 8, 9, 10, 15, 20, 50, and the final 144 pixels at the top are original (ie, camera sourced jpeg with text composited in and saved at quality 100).

The resulting image is 153k in size. In this example, a normal save of the source (captioned) image at a quality of 72 results in an image of equivalent size.

Speaking of size, since the cropped rows exist as images on their own, I can also see how much space each quality section uses.  So the q1 row (80pixels high – or about 8% of the final image) is just a tad over 3k in size – that is, about 2% of the bytes, and even the quality 50 stripe is only 8k – 5.5% of the bytes. However, the q100 stripe at the top – 144 pixels comprising 14% of the pixel space, and is a whopping 66% of the bytes!

In another example image, I discovered that you cannot drop a colour segment onto a greyscale image (but you CAN drop a greyscale segment onto a colour image, and then subsequently drop a colour segment within the greyscale. The resulting possum is shown, but I wont go into the making-of here.

Additional notes

In creating this, I found a few caveats…

This tree is terrible! It has a monotone border bits AND bad compression

• Make sure the crops and drops are aligned on the jpeg iMCU boundaries, or headaches ensue since the crop size will be silently altered in position (as, I think, is the drop). I made all my changes on the boundary of 16×16 blocks, though 8×8 is also possible (I believe this may depend on the jpeg)
• I had to set the caption text to a light grey after I found that pure white would cause jpeg errors when dropping (DCT coefficient out of range). I don’t know if this is a bug, or just a limitation in what’s possible due to jpeg encoding.
• A colour jpeg loses it’s colour when dropped onto a greyscale jpeg. Specifically:
  1. crop ‘cropped.jpg’ from colour source.jpg
  2. also create greyscale.jpg from source.jpg
  3. drop ‘cropped.jpg’ back onto the original location within greyscale.jpg to create merged.jpg
  4. merged.jpg is not a grey-with-colour-segment jpeg, but is all grey. in fact, it’s so grey that merged.jpg and greyscale.jpg are binary identical!
  • note: tested just once :)
• real-world practical applications of this? http://www.photopla.net/wwp0703/stripes.php …that’s probably about it. Any stylistic effect from this is likely achievable with less effort via a standard photo editor (The GIMP, etc) with the cost of a few % in filesize when saving at a high enough quality to ensure actual jpeg artifacts don’t spoil the effect.

This page is derived from my wiki notes on the subject, which include additional information/updates for the inquisitive.

See: http://wiki.thorx.net/wiki/JPEGhack

* create an m4a file (I made mine exactly 40seconds according to audacity and saved as . . . → Read More: my experiment in custom iPhone ringtones via linux]]> thanks to recent nautilus (I assume is responsible – as seen in Ubuntu Lucid Lynx) ability to view the iPhones file paths, we can try this (typing this as I go, so this is draft on-the-fly quality notes)

* create an m4a file (I made mine exactly 40seconds according to audacity and saved as a .wav, then used faac to convert to m4a (faac in.wav –title “ringtone experiment” -o tone.m4a ; mv tone.m4a tone.m4r)

* copy tone into iPhone://iTunes_Control/Ringtones/tone.m4r

* edit iPhone://iTunes_Control/iTunes/Ringtones.plist to add in the details of the new ringtone. Fortunately I had an existing one already, so I just copied that – updating the GUID (I changed one character), the Name (“funky new ringtone”) and the Total Time (40000) fields.

…and bingo, the ringtone is visible in the settings (selectable, plays), but does not lock in as the ringtone to use! ok, clearly some work needed.

incidentally, this is a 32GB iPhone 3Gs with iOS4.0.2, never been jailbroken or anything underhanded done to it before. It was unlocked from the carrier a few days ago (thanks .au law) and that required a lengthy restore, but otherwise smooth.

…ok, so back to it: plugged it into iTunes (Mac) – it couldn’t read from the device (an error!)

…rebooted phone – and the new ring tone now works correctly. YAY! =) is this success?

…plugging it into iTunes again and… it’s all “Verifying iPhone” at me – something I don’t generally recall. …ok, it’s backing up as normal now…

and the ringtone tab in iTunes tells me it doesn’t know about the tone I added in, it’s syncing, so I suspect it’ll delete my added tone…

…and checking in my iPhone after it’s done, what do you know – my custom added ringtone is all gone. Checking back via linux – and the tone.m4r file has been deleted, Ringtones.plist has been completely rewritten.

So, I assume (not surprisingly at all) that iTunes needs to know about it, as the authorative master, and tell the iPhone how it’s gonna be. This fits with my observed philosophy of the iPhone – as an adjunct to the master information on the computer.

…

so, round 2… I add the m4r into the iTunes library – it automatically adds it in as a ringtone – that’s nice. I assume iTunes assumed it was a ringtone by the file extension. And in the iPhone tab it shows up as available to be sync’d… and bingo. it works.

ok, so not a pure linux solution – unless you never sync your phone to iTunes anyway – but it DOES let you make a custom ringtone without requiring garageband. That’s something isn’t it?

And I assume the method would work under windows too… anyone want to give it a go and get back to me?

In fact, atq is literally out of order – I guess the idea is that you pipe it through sort. very very old-school . . . → Read More: building a better atq]]> I was frustrated at the poor information available by atq (what’s that? You’ll tell me WHEN you’re running stuff, but not WHAT you’re running? This seems a little out of order!)

In fact, atq is literally out of order – I guess the idea is that you pipe it through sort. very very old-school thinking

At the other end of the scale, at -c provides an information overload of EVERYTHING about the job. (60+ lines of env? yikes. I don’t need to know that most of the time!)

So where is the simple and useful summary that I’d genuinely expect of atq?
ie, not only the queued command times, but WHAT those commands are – one per line.

As a result, here is my simple “batq” – a better atq

#!/bin/bash

# a better atq
#
# returns the original atq output in bold
# AND IN ADDITION, ALSO:
# * returns items sorted
# * the intended working directory
# * the command to be run
#
        # TODO:
        # $command is not robust - will fail on multiple lines

function do_extraqinfo() {
    while read job ; do
        jobid=$(echo $job | awk '{print $1}')
        detail=$(at -c $jobid)
        dir=$(echo "$detail" | egrep ^cd | sed -e 's/cd \(.*\) || {/\1/g')
        command=$(echo "$detail" | tail -1)
        # alter the following line if the bold offends you
        tput bold ; echo -n "$job" ; tput sgr0
        echo "  $dir    $command"
    done
} 

atq | sort | do_extraqinfo

To summarise the previous posts…

The room: 14 – 30.5C

The rack: 23.3 – 43.4C

The room: 14.4 – 33.8C

The rack: 22.6 – . . . → Read More: as hot as… summer of ’10]]> So once again, my quarterly report on the temperature range outside my rack at knee level, and inside my rack at shoulder level – above two machines.

To summarise the previous posts…

Winter:

The room: 14 – 30.5C

The rack: 23.3 – 43.4C

Spring:

The room: 14.4 – 33.8C

The rack: 22.6 – 45C

And newly…

Summer:

The room: 19.6 – 34.9C
The rack: 27.9 – 45.7C

The seasonal trend is there. Not unexpected for it to be there, just nice to know what it is, quantitatively.

http://www.ted.com/talks/blaise_aguera.html

If you haven’t seen it, then do so now.

If you have, proceed…

So if I can be allowed to blow my own horn for a moment, I’d like to boast that I occasionally seem . . . → Read More: augmented thinking]]> Now this, I am sure you will agree, is a damn cool presentation of some excellent ideas.

http://www.ted.com/talks/blaise_aguera.html

If you haven’t seen it, then do so now.

If you have, proceed…

So if I can be allowed to blow my own horn for a moment, I’d like to boast that I occasionally seem to have quite marvelous ideas independently or others. Admittedly, others who were lucky enough to think of it earlier enough and/or had resources enough to make something of it. But my claim is that I sometimes think of these things _independently_.

And it has has been said of inventing – the best inventions are obvious, which is why it’s so humbling to remember that one didn’t think of these ‘obvious’ things themselves.

But in my case, can I finally “prove” that I did…?

Back to the video – I’d like to draw your attention to a point approximately 3:45 in: “…time travel. I’m not going to show you some of the wonderful historic imagery in here – there’s some in here with horses and carriages as well [...] it also is the foundation for augmented reality“.

Now compare with an email of mine to a friend, where I wrote this:

What would be kind of cool: augmented reality goggles (this is one of those ideas I had years before I knew there was a word for it) that show recreated historical versions of the world… but, instead of just showing a certain era, you see the ‘prettiest’ building at any given location. Making a best-of-all-possible-eras virtual world

So as far as I can tell (can anyone confirm?), this talk was made on Feb 11 or 12 (based on dated video comments afterwards). My email has a date stamp of: 10 Feb 2010 15:17:07 +1000). And, frankly, even if the video was recorded first, I sure didn’t see it till today

As an aside, my email was prompted by a rather ugly house I see regularly in town, and my own wonderings if it used to be pretty…

• • Contemporary Google maps street view
  • Sometime pre 1994, it was ugly then too
  • Circa 1930s – it was gorgeous.

I especially love how you get basically the same angle on all three – which was, of course, the key to my own inspiration

So clearly, the next thing for me to work out is real time travel – at which point I can begin to start sending these ideas of mine back to people in the past with the resources to make them a reality asap! =)

Here are the spring results

So in 3 months of spring, the ‘indoor’ (outside the rack) temperatures: 14.4 – 33.8C

And the “outdoor” (inside the rack) for the same winter season: 22.6 – 45C

So in short – not too different . . . → Read More: as hot as, end of spring edition]]> Previously I posted winter temperature ranges in the ThorxBlog post: as hot as…

Here are the spring results

So in 3 months of spring, the ‘indoor’ (outside the rack) temperatures: 14.4 – 33.8C

And the “outdoor” (inside the rack) for the same winter season: 22.6 – 45C

So in short – not too different from winter… a couple of degrees warmer all around… though I bet a bellcurve plot of common temperatures would show a slightly different bias to the story too.

Thermometer reset, and into summer… hello summer!

256 colour support is there. It’s in the code. It works. It’s nice. And . . . → Read More: Linux UI: crippling itself?]]> I was recently talking with a friend about commandline prompt gadgetry, and our tangent conversations drifted to 256colour support in X terminal emulators. xterm. gnome-terminal, and presumably kterm and others also, though only the first two have I looked at.

256 colour support is there. It’s in the code. It works. It’s nice. And it’s… not only not enabled by default, there is no clear and simple way of enabling it even if you WANTED TO.

I was going to blog more about this, but the following IRC fragment I think says it all. Any questions?

<nemo> it'd be sure nice for a distro to set itself up with no self-crippling
        for ancient stuff that, frankly, nobody is using anymore
<nemo> (or if they do, then it's being used by people savvy enough (probably)
        to work out the compatibility issues themselves
<nemo> ubuntu does a good step, I think, but probably not enough
<nemo> 256colours should be the GLOBAL DEFAULT these days
<Screwtape> Every time somebody suggests that, they get shouted down by people
        who say things like 'BUT WHAT HAPPENS WHEN I SSH TO MY
        FreeBSD/OpenBSD/Solaris/Dragonfly/Xenix MACHINE!'
<Screwtape> We can't even figure out whether backspace should be ^H or ^? 
<nemo> [a] don't use this distro then
<Screwtape> It's hardly global domination if everybody stops using it. 
<nemo> it's hardly global domination if we're compromising ourselves for the
        compatibility of those who DON'T use it
<nemo> [b] have a nifty 'set_maximum_compatibility' script, which backwardsify
        everything

...

<nemo> so, what else has potential, but is held back by... historical compatibility?
<nemo> and by 'potential', I mean "it's there on your system now, just waiting for
        you to apply the correct settings'
<nemo> I am identifying terminal colours, and default prompts
<nemo> (terminal colours, and all the terminals, applications (screen, elinks, etc),
        etc that go with.
<Screwtape> I.. can't think of anything else off the top of my head, sadly. 
<Screwtape> Or perhaps luckily.
...

In fairness, it’s not like it’s not been noticed before:

...
<nemo> https://bugzilla.gnome.org/show_bug.cgi?id=121800
<nemo> 2003... status new
<nemo> wtf
<nemo> https://bugzilla.gnome.org/show_bug.cgi?id=115750
<nemo> amuses me that my no-X11 server has terminfo for 'gnome' and 'gnome-256color',
        but my ubuntu desktop does not
...

And in case anyone wants to bring their terminal into the 21st century with 256colours instead of 8 (“tput colors” will tell you what yours can do), Screwtape has a nice little writeup here:

...
<Screwtape> http://screwtape.jottit.com/automatic_%24term_selection <-- Here's my
        automatic $TERM hack.

…

PS: As a final note: Shortly after this conversation, Screwtape filed the following debian bug report for vim’s syntax handling of sh scripts:  vim-runtime: sh syntax highlighting should default to POSIX.

…

PPS: Originally blogged to: http://blog.thorx.net/2009/10/linux-ui-crippling-itself. Please consider leaving comments at the blog rather than facebook or other social media side that this post may be propagated to. Thankyou