Martin Stut's IT Experiments

Blog has been moved to martinstutenglish.wordpress.com

2011-06-11T09:01:00.001+02:00

In order to consolidate my 3 separate blogs, I have moved all my English language blogs to http://martinstutenglish.wordpress.com. The content of this blog has been imported there, so you can replace your bookmarks and subscriptions

Blog has been moved to martinstutenglish.wordpress.com

2011-06-11T08:55:00.001+02:00

Comparing Functionality of Blogging Platforms

2011-06-02T16:13:00.001+02:00

With accounts at posterous.com, blogger.com and wordpress.com, I need to decide on a platform. One factor in deciding is the feature set of each platform, including the android client available.

Feature	Posterous	Blogger	Wordpress
automatic posting to Twitter, other blogs, social networks	yes, many services	not possible	a few services (Twitter, Faceboox, Yahoo Updates, Messenger Connect)
search engine positioning of your posts	poor	good	best
Creating a Draft on the Mobile	click the tiny "i" to mark the post as private. Later, when on the web, edit the post options (top right in a light gray, next to media) to make the post public.	click "Save as Draft"	set status as "Draft" before saving
Editing a Draft on the Mobile	not possible	straightforward	straightforward
Drafts when Editing on the Web	Save (don't publish yet) and then click Cancel. When warned about unsaved content, accept to leave the page.	Click "Save", not "Publish Post".	While composing, set the status to "Draft".
H1, H2 headings	Not offered in the online editor. You are able to enter these codes in raw HTML view. H1 becomes right adjusted - not looking nice.	Not offered in the online editor. You are able to enter these codes in raw HTML view. Looks good.	Offered in the online rich text editor. Looks good.
Tables	Not offered in the online editor. You are able to enter these codes in raw HTML view. Looks good.	Not offered in the online editor. You are able to enter these codes in raw HTML view. Looks good.	Not offered in the online editor. You are able to enter these codes in raw HTML view. Looks bad, because THs are silently removed.

Browsing the web for information, I stumbled across a blog post by novel writer Emlyn Chand: Which is better, WordPress or BlogSpot? See how they stack up here . The bottom line is: both have different strengths, it depends on your goal.

My Dive into the Social Web

2011-05-31T21:36:00.001+02:00

After reading about posterous, I decided to give it a try. My first impression is, well, impressed. For a first try, I set up a site, designed to be a single source for all my blogs - admittedly, some of them started as tests. In particular, the autopost function is nice to keep all the social channels updated.

Let's see whether I manage to write more now than I did before.

Connecting a Canon Pixma ip4600 Printer to Debian Squeeze Linux

2009-10-16T18:12:00.006+02:00

I bought that particular printer for its ability to directly print on CDs. The printer documentation says, that this needs the special driver (provided only for Windows), but still it should be able to print on standard paper from Linux.

Try 1: Plug & Play Connectivity

The Debian based Linux distributions nowadays do plug and play surprisingly well. So I gave it a try: connect the printer to a USB port, turn it on and see what happens:
A message box appeared "the ip4600 printer has been added with a text only driver". It offered "find driver", I accepted that offer, got asked for my password and voila, the printer configuration contained the ip4600_series.

Try 2: Manually Adding a Driver to CUPS

I connected my web browser to the local CUPS (https://localhost:631). The printer was there, but still with "generic text-only". Unfortunately, no search option turned out any results. So I looked into the aptitude package manager to install additional drivers.
I installed the package cups-driver-gutenprint, which pulled in ghostscript-cups.
The I chose the "add printer" dialog. Maker Canon, the model pixma ip4600 (gutenprint) was an offered choice which I chose. This created a new queue Canon_iP4600_series. I deleted the first automatic queue iP4600_series.

The first test print was a 6 page HTML file from Iceweasel (Firefox) which worked immediately.

Backup for Windows using a batch file

2009-10-02T21:23:00.002+02:00

This batch file does a reliable backup of Windows Vista Users to an external hard drive. There will be a new directory for each day the script is run. For Windows language settings other than German, you will need to adapt the SET FULLDATE line. FULLDATE should be something like 2009_10_31.

SET TARGETPATH=F:\Sicherung
SET FULLDATE=%date:~6,4%_%date:~3,2%_%date:~0,2%
echo FULLDATE=%FULLDATE%
SET TARGET=%TARGETPATH%\%FULLDATE%
mkdir %TARGET%
rem robocopy C:\Users %TARGET% /S /V /R:0 /W:0 /PURGE
xcopy C:\Users %TARGET% /S /C /I /R /Y

ToDo Lists GTD Style with TiddlyWiki, MonkeyGTD Style

2009-05-23T19:28:00.003+02:00

In one of my previous posts, I wrote about TiddlyWiki MPTW-Style. Back then in February I started using another TiddlyWiki variant, called MonkeyGTD, as a todo list structuring tool. I use Thinking Rock in the office and would like to have something different at home, to avoid being reminded too often of the huge backlog of work in the office.

MonkeyGTD uses the taggly tagging technique of MPTW to link tasks with projects, areas, realms etc. This way it is really easy to put your todo list into a structure. I really needed to read the documentation before starting, but I got rewarded by knowing how to arrange lots of tasks into imaginable chunks.

For my style of use, the major drawback of MonkeyGTD is the lack of sorting orders. With a dozen projects, each having several possible next actions, I'm overwhelmed by the quantity of possible next actions. I want a system that makes an educated suggestion of what to do next. MonkeyGTD has the option of starring projects and actions, but that is only a little help to me.

Data Recovery with Scrounge NTFS

2009-04-18T16:56:00.004+02:00

A friend asked me to try to recover data from an external 160 GB hard disk. It refused to show its contents on Windows XP, where it used to be used. In an earlier similar situation I had successfully used scrounge-ntfs. Scrounge NTFS is available as an out-of-the-box package for Debian Linux, which I run on my desktop computer at home. So I gave it a try in this case too. I connected the disk to a USB port of my computer, where it got the device ID /dev/sdb.

To use scrounge-ntfs, one needs some pieces of information (items copied from the author's website):

Start Sector: This is where the partition starts on your hard disk.
End Sector: Where your partition ends on the entire hard disk.
Cluster Size: This is the size of one 'block' of data on a partition
MFT Offset: The position of the NTFS Master File Table.

I didn't know this data, but I let scrounge-ntfs guess it: scrounge-ntfs -l /dev/sdb
I got
Start Sector End Sector Cluster Size MFT Offset
63 312579729 8 6291456
So I changed into the directory where I wanted the recovered data to end up and entered
scrounge-ntfs -m 6291456 -c 8 /dev/sdb 63 312579729
Unfortunately, scrounge-ntfs just complained about an invalid mft-record and stopped.
There is the option of running scrounge-ntfs without telling it where the MFT is. But if run so, it places all recovered files into one single directory. Manually sorting more than 20,000 files is not particularly funny, so I wanted to aim for a better result.

According to Stef Walter, the author of scroungs-ntfs, sector 6291456 is a customary place for the start of the MFT. Given the error message, I assumed that the usual place happened to be on a damaged sector. After a little thinking, I dared to use a trick: let it look for the MFT one sector after the original one, i.e. scrounge-ntfs -m 6291457 -c 8 /dev/sdb 63 312579729
I didn't really expect this to work, but it did. Although every message about a recovered file is preceded by scrounge-ntfs: invalid mft record, all files were recovered and by far most of them ended up in a meaningful subdirectory. So my friend is happy of having his data back again.

Personal Notetaking with TiddlyWiki, MPTW Style

2009-02-02T18:42:00.000+01:00

I developed the habit of trying to document everything I do. So I need a piece of software to quickly take notes and put them into relationships later. A wiki is likely to fill that need. While testing various server based wikis for my job (the largest branch is looking for a cantralized documentation system), I discovered an implementation I'd call an eye-opener: TiddlyWiki. The entire wiki is stored as a single, large (250 kB) HTML file. All the logic is implemented in JavaScript. I hadn't thought before that such an application could be feasible in JavaScript - that's why I call it an eye-opener.
Having all data in a single file and letting the browser run the logic makes it not well suited for group access, but ideal for personal use on a USB key. No installation required, just a decent web browser - optimized for Firefox, also working on IE. If you want to publish information contained in a TiddlyWiki, or just want a web hosting service for your private data, you can create a free account at tiddlyspot.com. I didn't try that, but many others have.

TiddlyWiki can be extended by plugins and these can be updated through the TiddlyWiki synchronisation feature. Several people have created such extensions, so there are many variants of TiddlyWiki around. All of them share the same core code.

After adding a few pages, I found it cumbersome to create the internal links. This is the reason, why I don't consider server based wiki an option for quick documentation writing (though wikis are great for documentation publishing and collaboration). Then I stumbled across a TiddlyWiki variant called MPTW. It contains a few function buttons which make life a lot easier for documentation:

On the top of each page (called "Tiddler" in TiddlyWiki parlance), all tags of this page are listed, together with a drop down menu, listing all other pages having this tag.
On the bottom of each page, there is a customizeable list of all pages having the title of current page as a tag. So I can create a page "VMware" containing an automatic listing of all other pages tagged with VMware.
On every page there is a "New here" button that creates a new page, automatically tagged with the title of the previous page. So the newly created page will automatically be listed at the bottom of the parent page. This way you can create a logical tree of pages without any extra tagging or linking effort.
The customizeable list of all child pages has an option ("sitemap") to display not only the direct children, but also the children of children and further generations. This mode really got me hooked.
The customizeable list of all child pages has an option to include an excerpt (one line or so) of each child page, so you see even more what's up in your documentation forest.

Try it and you'll love it.

Distributed Address Management with Zoho

2009-01-31T15:44:00.005+01:00

A friend of ours is a teacher and counsellor and needs to manage a collection of about 1000 addresses. He sends an email newsletter to about 500 of them every month and a paper mailing to about 200 of them twice a year. He asked my wife's office service to help him, so he can focus on spiritual tasks and doesn't need to spend much time for dry administration.
One option is to run a standard fat-client address management software (we're considering Optigem). But our friend lives about 15 km (10 miles) away, so driving there just to enter a single new address is not efficient. File server style access to an MS-Access database over an ADSL line (192 kbps upstream) or maintaining a Windows terminal server just for this purpose isn't either, so we'd have to copy the database to a USB key and physically exchange that in church on Sunday. This would imply, that between copies only one side is allowed to modify data. This would be acceptable, though not ideal. So I'm looking for a more elegant solution: a hosted address database on the web.

Looking for a Web Service: Zoho

There are some web services around, but the only free one I found to be able to manage categories was Zoho. They have a full CRM suite, but it was too big for this purpose. What thrilled me was Zoho personal (http://personal.zoho.com) - a complete work area with word processing, spreadsheet, mail etc. on the web. It does include a contact management component, separately accessible as http://contacts.zoho.com, with categories and permission groups. When editing a contact, you can add it to one or more categories and groups. Groups are for sharing data, so you could permit other Zoho users to see certain things in your workspace.
Sending an email message to an entire category is easy - but there is no Bcc option when composing messages. Bcc is a must, because you don't want your email address broadcast to many hundreds of other recipients of the newsletter.
So you need to export the contact data and compose the message in a conventional e-mail program like Thunderbird. This kind of export is possible for any single category. Formats offered include CSV, Outlook CSV and LDIF. LDIF is great for importing into Thunderbird as a new address book. To mail a newsletter in Thunderbird, you just need to enable the contact sidebar, select all addresses of the newly imported address book, right click on the selection and "add to Bcc".
With these possibilities, Zoho Personal is an acceptable choice as a contact management tool.

The only glitch I found was in localisation - my wife really prefers a German language user interface.
If you want the Zoho user interface in your language, as opposed to "English only", you need to login through personal.zoho.com, not through contacts.zoho.com. In personal.zoho.com, you can set a user interface language. But during the tests I could no longer get to the contacts application within the personal site. It just showed the form "please sign up". I sent a feedback to Zoho - and I'm curious whether there will be a response.

Magpie RSS: Include an RSS Feed on an HTML webpage

2009-01-24T17:18:00.023+01:00

Rationale for a PHP Feedreader

Quite often, organisational websites are written in static html, which is fine for the purpose, but want to run some kind of news system to publish current events without having to edit the core pages. If a CMS is available, one can use that for the purpose. But if they do not want to run a CMS - there may be good reasons, e.g. security - there is the option of using a blog service like blogger.com as a backend. But what to use as a frontend, visible to website visitors?
Telling people "please get a feedreader and subscribe to our blog" is creating an inadequate obstacle - many visitors of the target audience are barely capable of typing a URL into their browser, let alone manage bookmarks. So the feed needs to display automatically with the webpage. Including a Javascript feed reader would mean to require a client side script to pull data from a third party server (blogger.com), which is for good reasons blocked by default due to the "same origin policy".
So the feed needs to be included by a server side script. Nowadays PHP is available even with cheap shared hosting packages, so I could set out to create a PHP page blogfeed.php that pulls and displays the relevant items of the feed.

Magpie RSS

This task can't be unique, so I looked for a PHP script doing that. I found most refenrences for Magpie rss (http://magpierss.sourceforge.net/). This homepage also gives links to hints useful for RSS users and webmasters in general.
The mapie project does not offer a real manual, but it does provide links to examples.
The provided example only shows how to display a list of titles. My plan is to show the first three or so news items in full text and only headlines for the rest.
The author of http://www.hiash.com/blogger.php (referred to on the Magpie links page) has exactly the same plans a I did: using blogger.com as a backend for another webpage.
That made me confident enough to try it. At first I did it on my home (intranet) webserver:

My First Try with Magpie

I downloaded magpie in its most recent version 0.72 from the Sourceforge download page. There I discovered that the current version dates from 2005. I researched a bit for alternatives, but found only very few, very simple (probably too simple) scripts, e.g. LastRSS. So I concluded, that Magpie's age may be a sign of matureness and robustness, rather than obsoleteness.
All magpie users recommend this directory structure, so I set it up too:
magpie/rss_cache.inc
magpie/rss_fetch.inc
magpie/rss_parse.inc
magpie/rss_utils.inc
magpie/extlib/Snoopy.class.inc

The untarred download contains a nice file htdocs/cookbook.html.

For a first try, I created a little file http://www.home.stut.de/newsfeed.php (home intranet only) on the root directory of my home intranet webserver (apache2 with PHP5 on Debian Etch), copied almost literally from hiash.com:

define('MAGPIE_DIR', 'magpie/');
require_once(MAGPIE_DIR.'rss_fetch.inc');
$rss = fetch_rss( 'http://stut-tech.blogspot.com/feeds/posts/default' );
//display latest blog content:
$item = $rss->items[0];
$content = $item['atom_content'];
echo "<p>Latest Blog Entry:<br>$content</p>\n";

It really does show my last blog post. But I want to know more about the inner workings, so I try to look at the complete array contents using the print_r function:

Adding the newsfeed to an existing HTML-only website

I added a magpie folder (chmod 755) to the root directory of my website, and a blogfeed.php file to the root level of it.
I was able to call blogfeed.php in Firefox, but non-ASCII characters were broken. I solved this by adding a meta tag to the head section, so the charset became defined.


<html>
<head>
<title>Blogfeed on stut.de</title>
<meta http-equiv="Content-Type" content="text/html; charset=iso-8859-1">
</head>
<body>
<h2>Current Posts in my Tech-Blog</h2>
<?php
define('MAGPIE_DIR', 'magpie/');
define('MAGPIE_CACHE_ON', 1);
require_once(MAGPIE_DIR.'rss_fetch.inc');
$rss = fetch_rss( 'http://stut-tech.blogspot.com/feeds/posts/default' );
//display latest blog content:
//
$numberOfFullyDisplayedItems = 1;
$itemCounter = 0;
foreach ($rss->items as $item) {
$itemCounter += 1;
if ($itemCounter <= $numberOfFullyDisplayedItems) {
echo '<h3><a href="'.$item['link'].'" target="_blank">'.$item['title']."</a></h3>\n";
echo "<p>".$item['atom_content']."</p>\n";
}
// only for the first item after the completely displayed ones:
if ($itemCounter == $numberOfFullyDisplayedItems+1) {
echo "<h3>More Items</h3>\n";
echo "<ul>\n";
}
if ($itemCounter > $numberOfFullyDisplayedItems) {
echo '<li><a href="'.$item['link'].'" target="_blank">'.$item['title']."</a></li>\n";
}
}
// if there were more items: close the UL
if ($itemCounter > $numberOfFullyDisplayedItems) {
echo "</ul>\n";
}
?>
</body>
</html>

Now this code needs to get on the front page. The first obvious way would be to include it on index.php - but what if the start page is called index.html and changing the name to index.php is not an easy option?
Answer: Include blogfeed.php as an IFRAME. This is straightforward (if you are fluent in HTML and CSS):


<div id="rightbar" style="float:right;width:30%;">
<img src="mst.jpg" alt="Portrait of Martin Stut (JPG 5.5kB)" height="169" width="112" align="right" />
<iframe src="blogfeed.php" style="margin:2em 0em 1em 0em;width:100%;border:0px;height:80%;" />Current
  posts of my <a href="http://stut-tech.blogspot.com/" target="_blank">tech-blog</a></iframe>
</div>

You can view the result, together with a 1996 photo of me, at http://www.stut.de/martin.html

Create a Blog with a Subdomain of my own Domain

2009-01-24T15:48:00.004+01:00

All of this is already written in the online help of blogger.com, but I found it noteworthy how easy this is:

Log in to www.blogger.com and create a blog. For the time being assign it a-temporary-name.blogger.com.
Arrange a CNAME record for the (sub)domain of your future blog to point to ghs.google.com. I did that through the administration interface of my shared hosting account with Hetzner, so now orgblog.stut.de is a CNAME to ghs.google.com.
Because it was a new (sub-)domain of an existing domain, as opposed a new or changed domain, the entry spread across the Internet within less than a minute.
Enter your domain in the 'Custom Domain' option on the Settings | Publishing tab.

That's it. You can now look at my new brainchild using a stut.de address.

UltraVNC as offered by Heise (failed)

2009-01-24T15:38:00.004+01:00

The German IT publishing house Heise is offering a service to create an UltraVNC client that's customized to connect to your server: http://www.heise.de/netze/tools/fernwartung (German language only). Because I often need to help other computer users by remote access, the prospect of distributing a single exe file, which is able to establish a connection without installation, was appealing. So I decided to give the service a try.

I visited the URL, entered my home dyndns address and phone number, clicked on the download button and got the customized program as ctsupport.exe (c't ist the name of a German computer magazine, published by Heise). I transferred it to a virtual Windows 2000 machine. Of course I had setup my home router to forward incoming TCP connections on port 5500 to my client PC.

On the test machine, simulating someone in want of help, I double clicked the ctsupport.exe file, selected the single available connection (my server) and got connected - but only TCP-wise. The listener on my linux box did write about the connection, but did neither open a window nor write an error message. Apparently an UltraVNC server does not cooperate with any VNC-style viewer.
A glance at the Heise documentation says: "on Linux helper PCs use vncviewer". I had tightvncviewer, so that may have caused the problem. I noticed, that I already had vncviewer installed, so I simply tried vncviewer -listen on the helper PC, but the problem persisted: TCP connection established, but no windows opening and no error message on the client (helper).

At that point I decided to give up on the Heise tool and stick with "classic" TightVNC.

jbrout Photo Manager

2009-01-17T15:25:00.003+01:00

I'm no longer experimenting with this one, because I put it into continous "production" use. When I was looking for a solution in early 2008, jbrout was the only piece of software I found that fit my needs of managing the flood of picture my digital camera is producing:

graphical user interface, so adding and tagging pictures is quick and easy
use a published standard for storing tags and comments, so I can access the collection and the tagging information with a different program in five or ten years without re-tagging all of my 10.000 pictures.
I've learned this one the hard way, after 7 years of using a homegrown database dependent system that was too much effort to handle after returning with 800 pictures from a 3 week trip to Japan.
available for both Linux and Windows
free to use - I don't want to pay for software

jbrout is implemented in Python, a nice programming language that has runtime environments available for all major platforms.
It reads and writes the JPG comment fields of the picture file itself, so there are no external databases to synchronize. If you lose the few metadata files jbrout keeps, you can just re-import the picture directory (subdirectories are automatically included) and the index and the tag list is rebuilt automatically. The only thing you would lose is the categorization of tags - something I'm not even sure it is particularly useful, let alone required. I use tag categories, but start finding them irritating.

Tags are stored as IPTC tags. I've heard, that this standard is becoming obsolete, but I did not find any useable free software handling the new XMP standard. So I decided to stay with IPTC. When free XMP applications will come up, I hope they will have an option to import IPTC tags.

Getting started is as easy as it could be: install the software (platform dependent, but easy enough) and import the directories (called albums in jbrout) where you have already stored the pictures. You can have several albums in completely separate branches of your directory tree; all are accessible through a common index.
If there are already tags in the pictures, these tags are imported into the tag list. You can easily add new tags to the tag list.

Tagging images is straightforward: select the images to tag (even all the 800 pictures from Japan were o.k. to select in one go) and drag the tag (e.g. "Japan") onto one of the selected images. Depending on CPU and disk speed and the number of selected pictures, it might take a while until all pictures have got their tag.
Here are ballpark figures of the extremes I have encountered: The slowest, about 1 picture tag per second, was to a samba share on my 120 MHz P1 server (retired in May 2008) with a 40 GB IDE disk (bought in December 2001) over 54 Mbps WLAN. The fastest, about 10-20 picture tags per second, was locally on my 1.6 GHz P4 workhorse desktop with a new 500 GB IDE disk.

The search options are really cute. You can select single or multiple tags, time ranges etc., so if you got your tag system right, you can really find the buried gems in your picture collection.

Copy Audio CDs

2009-01-14T19:51:00.004+01:00

In the church I'm attending (Evangelische Gemeinschaft Marburg-Süd) we just started recording the sermon on CD in addition to cassette tape. These CDs are likely to be popular, so we'll need a method of duplicating them easily and quickly. I wanted to determine, how to do that on my PC (Debian Linux Testing, currently 5.0 "Lenny").

This may sound trivial to some people, but I had some special requirements:

Command line only, so I can create a script to fully automate the process.
There must not be any audible pause between tracks. Recording is done on an appliance type CD-recorder. Track marks are created by pushing the "record" button while a recording is under way. This is done live during the sermon and thus, more often than not, in the middle of a sentence - just after me noticing that a new logical section has begun.
We need multiple copies of a single master. Ripping the master CD-RW to computer files is permitted to take a while, but burning the copies afterwards should be fast.

I was suprised, how straightforward the solution is:

Ripping the Master CD-RW to files

Create the desired target directory and cd to it. It should be blank in order to accept files with a very generic name without conflict. I chose something like /backup/mr-sued/2009-01-11
Insert the master CD-RW into any suitable drive. I have a DVD-ROM and a CD-RW drive. I inserted the master CD into the DVD-ROM.
execute cdparanoia -B
No other parameters reqiured.

The audio tracks of the CD end up in the current directory with names track01.cdda.wav, track02.cdda.wav etc.

That's it.

Burning the Copies

Because burning CDs requires enhanced privilieges (write access to a device, assigning processes high priority in order to avoid buffer underruns), I did these steps as root. Burning as an ordinary user may or may not work, depending on your distribution and setup.

You need to know the Linux device name. I assigned /dev/cdrw to my CD-RW drive, so I knew I could use that. If you know less about your hardware, you can execute wodim -scanbus (some distributions might have cdrecord instead of wodim). This shows you a list of devices. Mine was

scsibus1000:
 1000,0,0 100000) *
 1000,1,0 100001) *
 1000,2,0 100002) 'HL-DT-ST' 'DVD-ROM GDR8160B' '0013' Removable CD-ROM
 1000,3,0 100003) 'AOPEN   ' 'CD-RW CRW3248   ' '1.10' Removable CD-ROM
 1000,4,0 100004) *
 1000,5,0 100005) *
 1000,6,0 100006) *
 1000,7,0 100007) *

The burning itself was easy, after an extended read of man wodim:

wodim dev=/dev/cdrw -dao -audio -copy -eject *.wav

Of course you should insert a blank CD-R (or a blanked CD-RW) before and you should be in the directory of the audio wav files.
If you are doing this for the first time on your computer, everybody recommends a test-run with an additional option -dummy.

The option -dao sets wodim in disk-at-once mode, as opposed to track-at-once. Disk-at-once ensures that there are no audible gaps between the tracks.
-audio makes sure you'll write audio tracks as opposed to data tracks. CD players can only play audio tracks.
-copy sets a permission bit to make arbitrary generations of digital copies of these tracks. We want to distribute the message. Your organisational situation may be different.
-eject is optional fun, but somewhat comfortable: eject the finished CD after it has been burnt.

Happy Hacking!