Archive | Computing RSS for this section

TDE icon stuff — extract an icon from an EXE or DLL

First, I got a look at the old TDE icon — you can use icoutils to extract icons from Windows EXE and DLL files:

$ apt-cyg install icoutils
$ wrestool tdew.exe
--type=3 --name=1 --language=1033 [type=icon offset=0x580ec size=872]
--type=14 --name=1 --language=1033 [type=group_icon offset=0x58458 size=20]
--type=16 --name=1 --language=1033 [type=version offset=0x58470 size=752]
$ wrestool -x -t14 --output=. tdew.exe



Very small, quite pixelly.

Fuzzy T

Old TDE icon

$ convert tdew.exe_14_1.ico tdew.exe_14_1.png

To get a replacement that matches, I imported this into a drawing program, even MSPAINT.EXE would do, scaled it up, typed a capital T in a matching font over the top of it (bold Times New Roman seems to match closely), then made a new, empty file and used the same font, and and saved it back to an ico file (can use convert to make the ico).

New one is just a white T on a black background, with rounded corners, cutoff bits transparent (see eg this post).

Here we have it:


New less fuzzy icon

Bigger, possibly better …



Wrong tab icons in Firefox

Tabs in Firefox show the wrong icons.

Mismatched icons -- eg ebay has Telstra icon

Mismatched icons crossed out!

I opened a terminal and typed:

$ find . -iname "favicon*"

And it dug up:


So I quite Firefox and went:

$ rm ./.mozilla/firefox/abcdef.default-1234567890/favicons.sqlite

Then restarted Firefox.

I now got:

All blank icons

Is this any better?

Now, I visited each link.

Most of them found and used the correct icon. A few did not; they mostly sorted themselves out soon enough! Better, I think.


The twin window manager

The Text WINdow manager, twin, is quite a nice piece of software if you don’t
want to have to run X, though it can run within an X term.

GitHub – cosmos72/twin: Text mode window environment. A “retro” program for embedded or remote systems, that doubles as X11 terminal and text-mode equivalent of VNC server

I have found that it works extremely well with the ‘leggie’ fonts, from
Leggie, a legible, pretty bitmap font

Here is an example of a twin session, captured using fbcat.

4 windows within a twin session

twin is great for running a bunch of text applications

So this was a twin session, using the leggie18 font, on a netbook.

Top-left is a Gopherus session, top right is a window on my account, bottom-left is Vim editing a LaTeX file and bottom-right is an Alpine
mail window.

twin has many capabilities — this only shows you what it looks like. One cannot run framebuffer graphics inside twin windows — eg fbi, or dosbox (which uses sdl) — but I have found that a twin window on one virtual console, and then a ‘bare’ framebuffer on a second one makes for a good combination for working without X.

I have noticed that some fonts give ugly outlines on the windows (rows of diamonds or non-characters). The leggie fonts on 32-bit Debian give neat lines around the windows. YMMV.

Some nice features include a built-in clock (see bottom-right corner), and the ability to type in a window while keeping it behind others. That’s why the screen capture command is invisible in the screen grab above. Alpine, links, lynx and other text-based network tools work fine, plus all your other console tools. It is very light on resources, too. top suggests it uses about 0.5% of my memory and CPU — and I am running an old netbook with 1GB RAM and an Atom N550 chip!

Could form the basis of a Linux distro to challenge TinyCore,



Black screen with mouse after Windows login

Note to self

  1. Booted Win 10 machine.
  2. Got usual login screen.
  3. Logged in as per usual.
  4. Got a black screen with a working, visible mouse pointer.
  5. Ctrl-Alt-Del.
  6. Got the menu and chose Task Manager.
  7. Went to start up items and disabled most items, but enabled ‘Lenovo Utility’.
  8. Used Ctrl-Alt-Del to reboot.
  9. Got the screen back.

Of course, I’ve skipped all the messing around that I really did.

I don’t know why. Other non-admin account on the machine worked fine, so something funny happened in a setting somewhere. Anyway, there you go.

Win 10

tilde — my crude automated webpage

The whole world of ’tilde’ accounts is to give you a little server space where you can monkey around in bash and maybe build a website the 1990s way — by editing HTML in nano, emacs or vim.

Mine is here:, and this is how it looked as-provided by the admin:

A brand new tilde club page

As provided

I have set it up so that if I scp files into the right directory, they will be automatically added to the index.html.

Consider the following bash script. I log into my account and run it from ~/public_html using nohup so that it does not stop when I log out. The script lives in the ~/public_html/pages directory and just creates a block of HTML with links to each of the HTML files it finds in there.

The script then creates a little file with the time and date updated information in it, then cats all the bits together, then sleeps for a day and runs again. My scripting is very crude, but seems to work.

while true
  cd pages
  cd  ..
  echo \<p\> \</p\> > date
  echo \<p\> \</p\> >> date
  echo Last updated: `date` >> date
  cat pages/index.middle.html date index.tail.html > index.html
  rm date
  sleep 24h

Here is the text of It loops over all the .html files in the directory. The echo command prints out a line of the form of a basic HTML link. It grabs the second line of the html file for use as the link text (that is what the head/tail bit does) then creates the link.

rm index.middle.html
for f in *.html
	echo \<p\>\<a href=\"pages/$f\"\>$(head -2 $f | tail -1)\</a\>\</p\> >> index.middle.html

The output of looks something like

$ cat pages/index.middle.html 
<p><a href="pages/hermes10.html">The Hermes 10 electric typewriter</a></p>

So all I have to do is make sure I upload a HTML file (with its dependencies) that looks something like the one below. Top 3 lines are:

  1. open comment
  2. link text
  3. close comment.

After that, any legit HTML should do.

Link text - always here.
  <h1>Main heading</h1>



So I make up my new page locally, with the correct 3-line header, then scp it to the correct folder, and the script wakes up once a day and adds the page to the index.

Crude, but effective. Obviously, I can complexify what I do and add more features, but I’ll let that evolve with time.



Simple surface plot in gnuplot

Here’s a random example for no good reason.

Here is the gnuplot script (

set iso 30
set samp 50
unset key
#set title "sin(r)"
set xlabel "x" font "Times:Italic,14"
set ylabel "y" font "Times:Italic,14"
set zlabel "z" font "Times:Italic,14"
set xrange [-4:4]
set yrange [-4:4]
set xtics offset -0.5,-0.5
set ztics 1
unset surf
set style line 1 lt 4 lw 0.5
set pm3d
set term post level1 color font "Times,12" fontscale 1.0
set output "plotfile.eps"
splot sin(sqrt(x**2+y**2))

Here are the commands run at the command line:

$ gnuplot 
$ epspdf plotfile.eps
$ xpdf plotfile.pdf
$ pdftoppm.exe -r 600 plotfile.pdf > plotfile.ppm
$ convert plotfile.ppm plotfile.png
$ display plotfile.png
$ rm plotfile.ppm

And this gives me an eps, a pdf and a png:

89K  plotfile.eps
56K  plotfile.pdf
990K plotfile.png
A coloured surface plot drawn in gnuplot using the script above


And here’s a simple script to plot sections through the surface:

$ cat 
unset key
#set title "sin(r)"
set xlabel "x" font "Times:Italic,18"
set ylabel "z" font "Times:Italic,18"
set xrange [-4:4]
set yrange [-1:1]
set border lw 0.25
#set style line 1 lt 4 lw 0.5
set term post level1 color font "Times,12" fontscale 1.0
set output "plotfile-cut-y=0.eps"
plot sin(sqrt(x**2+0**2)) lc rgb 'black' lw 4
set output "plotfile-cut-y=1.eps"
plot sin(sqrt(x**2+1**2)) lc rgb 'black' lw 4

And here is plotfile-cut-y=1

A section through the surface, where y=1; looks a bit like a capital M

Section where y=1


Making a little icon for the webpage

You know those icons you get in the tabs when you view a webpage? They are really easy to do.

This is how I did one for my page.

First, I found an image I wanted (from something related to this). It was in a PDF I had made some time ago, so I converted the PDF to a raster format, then cut out the bit wanted.

$ pdftoppm -f 1 -l 1 -r 600 mzfntdoc.pdf > icn.ppm
$ pnmcut 2935 2600 138 80 icn.ppm > cut.pnm

Convert to ico and png

$ convert cut.ppm cut.png
$ convert cut.ppm cut.ico

Copy to the filespace on the account.

$ scp cut.png

Add a single line to the index.html — in the <head> … </head> field.

	<title>Random pages</title>
	<link rel="icon" href="cut.png">

That’s it. Here is the icon:

M Z in blocky text

The icon

And here it is in action — the webpage viewed in Firefox shows the icon in the tab.

The icon appears on the tab in Firefox

There it is!

That was easy!

Some random BibTeX bibliography style examples

Here we have some basic BibTeX bibliography styles, with a few of the key distinguishing features pointed out.

First, good old unsrt.bst

an unsorted bibliography style

Then, chicago (from natbib)

chicago manual of style

The Harvard-ish agsm style

the agsm style, derived from Harvard -- author -- year

And, last, an APA-based style (apalike, from natbib)

another author/year style


Well howdy

Text News — news in plain text

Recently I got a account. The account comes with 2 directories — public_html and public_gopher. So I started messing around with gopher, using the gopher client and the gopherus client (and lynx works too).

One of the most useful gopher pages I found was Text News. It sucks down RSS feeds and presents them as nicely formatted plain text.

Very nice, but as an Australian I’d like to read some Australian news. At present I’m not going to make my own gopher page or anything like that, though it would be possible to use the gopher server on tilde to do that, and I might (and/or html).

Very kindly, the Text News page explains how to use the script that grabs HTML and RSS pages and formats them for plain text reading.

So …

  1. Using gopher, visited Text News on gopher and then downloaded the text file that describes how it works; saved to …
  2. $ mkdir installs/rsstotext
    $ mv instructions.txt installs/rsstotext/
    $ cd installs/rsstotext
    $ cat instructions.txt
  3. Viewed the file and installed stuff:
    $ sudo apt-get install python python-pip #python2
    $ sudo pip install html2text requests readability-lxml feedparser
    $ git clone
  4. Went to find some feeds
    $ links2 -g
  5. Searched for a list of Aussie RSS news feeds; found one at:
    Downloaded as ausfeeds.html
  6. Viewed the file and worked out how to most easily pull out the URLs of the feeds; don’t mind if it is a little bit manual.
    $ grep xml ausfeeds.html | cut -d'=' -f2 > ausfeeds
  7. Checked usage:
    $ cat instructions.txt
  8. Edited the resulting file, including prepending the appropriate command to each line, and commenting out the ones I don’t want just now.
    $ vim ausfeeds
    $ cat ausfeeds
    #! /bin/bash
    echo "Consider cleaning up in  /home/username/installs/rsstotext/saved!"
    cd /home/username/installs/rsstotext
    echo ABC ...
    python /home/username/installs/rsstotext/ --rss -n --url
    #echo SMH ...
    #python /home/username/installs/rsstotext/ --rss -n --url
    echo Age ...
    python /home/username/installs/rsstotext/ --rss -n --url
    echo Huffington Post Australia ...
    python /home/username/installs/rsstotext/ --rss -n --url
    echo Canberra Times ...
    python /home/username/installs/rsstotext/ --rss -n --url
    #echo WA Today ...
    #python /home/username/installs/rsstotext/ --rss -n --url
    #echo Brisbane Times ...
    #python /home/username/installs/rsstotext/ --rss -n --url
    echo Done! News stored in /home/username/installs/rsstotext/saved
    sleep 2s
    cd saved
  9. Made it executable (later made a soft link into ~/bin):
    $ chmod +x ausfeeds
  10. Try it…
    $ ./ausfeeds

It takes a while, but works fine. A text mode file manager is a good way to view the results; Here we use vfu; here is what it looks like.

A news feed viewed as a formatted text file

Viewed in vfu

Once the directories are populated, it’s best to not rerun the script until an updated list of stories is desired — just use vfu to browse the existing downloads.

Gopher it!

NetComm-M2M NTC-40WV Firmware Update

We’ve been having a connection problem with this router — a computer on a wireless connection will often find the router but not the internet, though it worked on last login. Any wired connections can see the interwebs. So I guess that’s a ‘wireless connection problem’.

Power cycling (have you tried turning it off and on again?) the router solves the problem, but is not a very satisfactory solution.

So I thought I’d try a firmware upgrade.

Old firmware was 1.9.1088 (or something), quite old. So in a way that was good — more chance that the new firmware will solve the problem.

Went to and downloaded This is probably the last firmware this (now unsupported) device will receive.

Unpacked the archive.

Logged into router ( as root.

Now, the main trick is that when upgrading from 1.x.x.x series firmware to 2.x.x.x, we must first install something called appweb-large-file-

This is within the downloaded archive, so it’s OK. Don’t go hunting around in the ‘net for it.

In the browser interface to the router, went to System → Load/Save → Settings and saved my existing router settings to a backup. (Accept the given file name!)

Then went to System → Load/Save → Upload and browsed for the ipk file, then uploaded and clicked ‘Install’. (The instructions are in a PDF that’s also in the archive; read that instead of this.)

The 2.x.x.x series firmware is a single big file, where earlier versions contained two smaller files, hence the need for the preliminary upload.

Then browsed again and chose ntc_40wv_2.0.27.3.cdi.

Uploaded and installed.

Waited about 4 or 5 minutes.

Shiny new status page and rearranged browser interface to the router appeared. Was active; I could browse the web etc, so all looked good. But — had our problems with wireless connections gone away?

<<One week later>>

Now, after a week of observing the ability of machines to connect, do we still need to reboot the router regularly?

It seems … a lot better!