Qxw and Crux: Frequently Asked Questions

What sort of grid can I construct with Qxw?

You can construct grids of any shape based on square or hexagonal cells or based on a circular design. You can use blocks or bars, or a mixture of the two, to separate words. The maximum grid dimension is 63 and the maximum length of a light is 250 characters.

What languages does Qxw support?

Release 20200708 of Qxw has built-in support for the following languages: Czech, Danish, Dutch, English, Estonian, Finnish, French and Italian, German, Ancient Greek (polytonic), Modern Greek (monotonic), Hungarian, Latin, Norwegian, Polish, Romanian, Russian, Slovenian, Spanish and Swedish. In addition, you can create customised alphabets to support other languages.

Qxw’s user interface is currently implemented in English only.

What dictionary does Qxw use?

In the Linux version, if no dictionary is specified using the command-line -d option, Qxw looks for the following dictionary files and loads the first one it finds:

• the file /usr/dict/words
• the file /usr/share/dict/words
• the file /usr/share/dict/british-english
• the file /usr/share/dict/american-english

Note that on some Linux distributions, the last of these is a version of the Moby word list. This is too comprehensive for many purposes, including many strings of characters that are not obviously words at all.

The Windows version comes bundled with a copy of the UKACD (see below).

You can make Qxw use a different dictionary file using the menu system or with a command-line argument: qxw -d<filename>. The argument can be repeated if several dictionaries are to be loaded.

The filenames of the dictionaries used in constructing a crossword are saved along with the grid.

Crux’s default dictionary is the file dict in the directory from which it is run.

What is the format of Qxw’s dictionary file?

The dictionary is a plain text file, which means you can use standard text tools to create and edit dictionaries. Each line of the file should contain one word.

Qxw first assumes that the dictionary file is encoded using one of the UTF-8, UTF-16 (big- or little-endian) or UTF-32 (big- or little-endian) formats. If when reading the file it determines that this is not the case, it tries again on the assumption that the file is encoded in ISO 8859-1 format.

If you are using the default Roman alphabet configuration, simple puncutation marks are stripped from words; ‘words’ containing only non-alphanumeric characters are ignored entirely. Accents are removed and all letters are converted to upper-case. Other alphabet configurations behave differently, in particular with regard to the treatment of accents.

Words may contain spaces, which are removed.

Where can I find alternative dictionary files?

There are many freely-available word lists to be found on the Internet. Here are a couple of places to start looking:

• UKACD18+, a version of UKACD18 with the bowdlerisations from UKACD17 undone and some errors corrected
• UKACD18
• UKACD17
• Moby, an enormous public domain word list
• YAWL (Yet Another Word List)
• The Sourceforge word list page has many useful links

For licensing information see here.

Do you have a version of Qxw for DOS/Windows 3.1/95/98/Me/NT/2000/XP/Solaris/Arduino/PIC16F628...?

Release 20140331 of Qxw runs under Windows XP and Windows Vista. There are Debian packages available for various architectures, although not always for the most recent release: see https://packages.debian.org/search?keywords=qxw. Otherwise no, sorry. It should be possible to run Qxw in a virtual machine; it appears to run well under Wine.

... for Raspberry Pi, perhaps?

Qxw can be built from source on a Raspberry Pi.

... how about Mac OS X?

Dan Friedman reports success building and running Qxw under Mac OS X Yosemite (version 10.10) and El Capitan (version 10.11). His instructions can be found here.

Alternatively, you may be able to run Qxw in a virtual machine under Mac OS X, or using Parallels or Wine.

Why can’t I get your program to work under my distribution of Linux?

The binaries distributed are compiled for the Intel x86 architecture. These will only work on Intel and compatible (for example AMD) processors. You need reasonably up-to-date versions of GLib and GTK: try building from the source code on your own machine if that seems to be the problem. Otherwise, please contact me and I will try to help.

How do I convert an EPS or SVG file into GIF or PNG format so I can publish it on a website?

Note that Qxw can export grids directly in PNG format or (square grids only) in HTML.

The easiest approach is to use Inkscape, which is available for both Linux and Windows.

Why does the HTML I have exported from Qxw not display correctly?

The HTML that Qxw produces requires that the browser support CSS (Cascading Style Sheets). Some older browsers do not support CSS fully. Many browsers print pages slightly differently from the way they appear on the screen, and some early versions of Firefox are rather poor in this regard.

Earlier versions of Qxw emit HTML that refers to two small image files, qpx0.gif and qpx1.gif, which were included with Qxw. You must ensure that there are copies of these files in the same directory as the HTML file. These files are not required for HTML produced by the current version of Qxw.

If you still experience problems with the compatibility of the HTML output, please contact me with the details. In particular I shall need to know which version of which browser you are using.

How does the grid filler work?

Qxw’s grid filler and Crux work in very similar ways. The essential idea is to explore the possible fills of the grid one letter at a time, trying to solve the trickiest parts of the grid first. When a letter is hypothesised for a particular cell, the implications of that choice are worked out and propagated around the grid. This is used to determine which cell is tried next.

What do Crux’s progress messages mean?

The counts show the number of different letters tried at each search depth.

What is the format of Crux’s dictionary file?

The dictionary file should contain one word per line. Only the characters ‘A’ to ‘Z’ and ‘a’ to ‘z’ are significant; Crux is not case sensitive in the dictionary file. White space separates words. If you wish to use a dictionary that contains phrases, any spaces between the words of a phrase should be removed first.