A proposed new language construct: do/undo

There are several cases, in which a program has to do operations A,B,C. If an error occurs while doing A, the program needs to undo A. If there is an error while doing B, the program needs to undo B then A. Similarly, an error in C requires undoing of C,B,A in this sequence.
Continue reading “A proposed new language construct: do/undo”

A better way to run Android Studio from a Docker container

Did you want to build a Docker container, which will run Android Studio for you, for example because your PC does not run GNU/Linux Ubuntu 14.04 but another Linux distribution?

You can find suitable scripts in the GitHub repository tddpirate/androiddocker

האם רצית לבנות Docker container שיריץ לך את Android Studio, למשל מפני שמחשבך אינו מריץ GNU/Linux Ubuntu 14.04 אלא הפצת לינוקס אחרת?

ניתן למצוא סקריפטים מתאימים בגיטהאב: tddpirate/androiddocker

The Python module for file type identification, called ‘magic’, is not standardized

I found the hard way that the API exported by the Python module ‘magic’ differs among different versions of the module.

The version installed when installing the Debian package ‘python-magic’ expects the following API:

import magic
mymagic = magic.open(magic.MAGIC_MIME_TYPE)
mymagic.load()
mtype = mymagic.file(inpfname)
print("The MIME type of the file %s is %s" % (inpfname,mtype))

The version installed using ‘pip install python-magic’ expects the following API:

import magic
mymagic = magic.Magic(mime=True)
mtype = mymagic.from_file(inpfname)
print("The MIME type of the file %s is %s" % (inpfname,mtype))

The following code allows the rest of the script to work the same way with either version of ‘magic’:

import magic
def build_magic():
  try:
    mymagic = magic.open(magic.MAGIC_MIME_TYPE)
    mymagic.load()
  except AttributeError,e:
    mymagic = magic.Magic(mime=True)
    mymagic.file = mymagic.from_file
  return(mymagic)
mymagic = build_magic()
mtype = mymagic.file(inpfname)
print("The MIME type of the file %s is %s" % (inpfname,mtype))

Learn to develop for the Android, contribute to Free Software and advertise yourself

During the last few weeks I developed an Android application for helping people prepare for the Israeli driving theory test.  The application uses the questions database available from http://data.gov.il/dataset/249 (and the corresponding ones for Russian and Arabic).  It was a nice and fun exercise, which helped me master more parts of the Android development platform.

The application is now at version 0.1 and is working.  Its look and feel is not polished, yet; and some missing features have been identified by comparing it to three other applications available from Google Play doing the same thing.

Today the source code has been released under the GPLv3 and is available from https://github.com/tddpirate/teuria and if you are looking for an Android developer with few weeks worth of experience and known quality of work, you may have a look at the above.

Installing Slackware (a Linux distribution) in a VirtualBox virtual machine

After not having dealt with Slackware for several years, I needed to create a virtual machine with Slackware.  I had to solve two problems, and am writing below how they were solved for the benefit of future Slackware based virtual machine builders.

  • Slackware version: 13.37 for 32-bit x86, installed from DVD
  • VirtualBox version: 4.1.18 r78361, Debian Squeeze amd64

Virtual machine parameters:

  • Operating system: Linux
  • Version: Linux 2.6 (32-bit being implied) – the combobox list had no entry for Slackware.
  • Base memory: 1.5GB
  • Hard disk: 30GB
  • Host I/O caching: on (to silence a warning).

Problem 1
When trying to boot from the DVD in order to install Slackware 13.37, the hugesmp.s kernel failed to boot.
I found the following bug reports:
https://bugs.launchpad.net/ubuntu/+source/linux/+bug/246067
http://www.linuxquestions.org/questions/slackware-14/kernel-panic-on-virtual-box-installation-its-driving-me-nuts-901223/
https://forums.virtualbox.org/viewtopic.php?f=3&t=44598

Some of the above bug reports had the suggestion to turn on VT-x/AMD-V acceleration.  However it was already turned on in my virtual machine.  I tried several combinations of flags in Settings/System/Processor and Settings/System/Acceleration but of no avail.

The problem was solved by changing the IDE controller type from PIIX4 to PIIX3 in Settings/Storage/IDE Controller/Type.

Problem 2
The initial reboot after installation failed.  From the stack trace it seemed that the problem was due to unknown kernel options.  I found some bug reports, but they were not directly related to this.

I successfully rebooted the virtual machine when entering the following as LILO prompt:

Linux pci=routeirq acpi=off pci=noacpi

Then, the problem was solved by editing /etc/lilo.conf as follows.  The line

append=” vt.default_utf8=0″

was changed to

append=” vt.default_utf8=0 pci=routeirq acpi=off pci=noacpi”

And then /sbin/lilo was run (without arguments).

Computerized elections in Israel – the threat is still there

Time has passed since the Nov. 2007 council elections (with no publicity to the results of the pilot of computerized elections), and since the Dec. 2008 Taldor fiasco.

And it turns out that in their infinite wisdom, the Israeli government decided to continue to cater to the impatient, “want to know the results right NOW!!!” segment of the electorate in future elections.  Having learned from the experience of other countries, a system based upon near-field RFID tags was chosen for development and deployment.

However, according to http://www.eng.tau.ac.il/%7Eyash/evoting-relay-rfid2010.pdf, the chosen system suffers from serious vulnerabilities.

It seems that the real agenda of the politicians here, like everywhere, is still to get the populace to accept ballot systems, which allow the elections to be rigged.

Web 2.0 Best Practices

Suppose you have been contacted by someone with a great Web 2.0 idea and he wants you to join his startup.
You need to know whether he knows what he is talking about.
The following checklist may help you tell the clueful apart from clueless.
I hope people will be able to contribute advice concerning each item in the checklist as well as more items I missed.

  1. Dealing with bad content:
    • Spam
    • Trolling
    • Off-topic user-contributed content
  2. Vandalism (and in general, content backup/restore).
  3. Legal:
    • Acceptable use guidelines
    • Copyright violations and other issues
  4. Content ownership/lockdown policies – will a contributor be able to export his contributions into file/s in his own PC?
  5. How will the network effect be overcome (if another Web 2.0 site already exists serving the same need, how to get people to use your Web site instead of the other site, if they already have stuff).
  6. Business model (i.e. how to actually get people to pay for the stuff).
  7. Scaling with demand (nowadays, thanks to cloud computing services availability, the required scaling is not that of servers but that of customer service personnel and maybe other critical resources).
  8. Are there standards (such as XML schema) relevant to the kind of content to be served by the site?

The hospital which demands that its surgeons operate in non-sterile theaters, with inadequate equipment and without enough help

If what Alan Carter says in his The Programmers’ Stone blog is right, then the way our society treats software developers is like requiring surgeons to operate in non-sterile theaters, with inadequate equipment and without enough help from other doctors and nurses.

Thank you, Taldor LTD., for tarnishing the reputation of computerized elections in Israel

Yesterday, the primaries election of the Labor party in Israel was aborted and postponed to a later date due to serious malfunctioning of the voting machines deployed for this purpose.

The provider of those machines was Taldor LTD.

Sources (in Hebrew):

I wrote previously about computerized elections.

Copyfree vs. Copyright/Copyleft

The other day I stumbled upon the Website http://copyfree.org/, which advocates a software licensing model somewhat similar to LGPL.
See the Website for arguments in favor of this licensing model.

I would, however, stick with GPL/LGPL due to the following reasons:

  1. The world has some actors (such as the one whose name starts with M and ends with T) with monopolistic intentions. Copyfree is not strong enough to stop them. GPL (especially its v3) is essential to limit the effects of such actors.
  2. Some software developers are not altruistic philanthropists. They expect to be compensated for their software development work. In the case of software which scratches their own itches, an acceptable form of compensation would be enhancements to the software, which fix bugs and – more interestingly – add new features. When wielded by such developers, GPL/LGPL are used much as traditional copyright law is used by creators to get compensated for their creations.
  3. In the special case of security software, which should be used by everyone, exemptions can be made on case by case basis. The reasoning is much the same as the one which led USA to release to USSR, in midst of the Cold War, certain technologies for securing atom bombs against accidental detonation. And those were days, in which people were executed for releasing nuclear secrets to the wrong parties (witness the Rosenbergs affair).