Announcing the return of WssIndex

From the early 1980's to the mid 1990's, WssIndex was the premier disk cataloging program for DOS. Under Windows, it ran only as a console mode application. A GUI version has been in the pipeline longer than Duke Nukem Forever, and now it's actually being released. The Qt Toolkit by Nokia provides the graphical interface.

Licensing

WssIndex 8 is released as GPL 3.0 freeware.
(7 was the highest version number for console-mode versions of WssIndex. These versions are hopelessly obsolete and no longer supported.)

System requirements

Windows XP or higher.
Windows 2000 seems to work, but since this OS is not officially supported by Qt (or, for that matter, by Microsoft), continued compatibility cannot be guaranteed.
Older versions of Windows will likely yield unresolvable missing DLL errors.
64-bit version requires 64-bit Windows.

Two 32-bit versions of WssIndex are offered, one built using the MinGW (Minimalist Gnu for Windows) C++ compiler, the other using Microsoft Visual C++ 2008 Express. They look and operate identically, but they need different DLLs. (Appropriate DLLs are included with the installers. The DLLs are installed in the same directory as the program, so both versions of WssIndex, and other Qt-based programs, can co-exist.) Performance differences are minimal. The VC++ version indexes my C drive roughly 5% faster. On the other hand, either version speeds up by a factor of more than two if the disk is indexed a second time, presumably because the operating system has cached directory information.

A version for 64-bit Windows, built with Visual C++ 2010, is also available. The 64-bit version is not subject to the 2 GB process space limit imposed by 32-bit Windows so larger databases are be possible. You'll need 4 or more GB of physical memory to take advantage of the increased capacity.

Discussion Forums

Discuss WssIndex in the Forums. Sorry if you see spam here. I clean up regularly, but the low-lifes create new accounts and post more.

Screen shots

Indexing under Windows 7

Changelog

Version 8 update history

Downloads

Note: Norton Antivirus dowload insight doesn't like files it has never seen before. Download of beta 1 was blocked until their whitelisting process was completed; now you just get a warning that the file is unknown. Future releases will probably be treated the same way. If you're worried, you can submit the download URLs to VirusTotal which will download the installer and scan it with 43 different scanners.

WssIndex 8.0 installer package, MinGW version Executable with needed Qt 4.72 and MinGW DLLs.
WssIndex 8.0 installer package, Visual C++ 2008 version Executable with needed Qt 4.72 DLLs.
WssIndex 8.0 installer package, 64-bit Visual C++ 2010 version Installer will download redistributable package if needed.
Microsoft Visual C++ 2008 Redistributable Package This is only needed for the 32-bit Visual C++ version of WssIndex on Vista or older. Many programs use VC++ 2008 so you probably already have this installed. If you get errors about missing msvcp90.dll or mscvr90.dll, you probably do need to install this.
Windows Installer 3.0 Redistributable Again, this is something you probably already have. Install only if you are unable to install the VC++ 2008 Redistributable package above.
WssIndex 8.0 source code Only needed if you want to build from source.
Qt by Nokia Only needed if you want to build from source.

Support for non-Windows operating systems

One reason for choosing Qt for WssIndex was its support for multiple platforms. Early on, the Qt-based version of WssIndex was only a viewer without Unicode support. At this stage, I copied the source to a Fedora 12 Linux system. After a few trivial code changes, it compiled and worked. But, there are some serious obstacles to ports to other systems:

Drive letters: The concept of drive letters is unknown to most operating systems, so indexing by drive letter doesn't make sense. Indexing by UNC path could be done instead.
File extensions: Not all operating systems use file extensions to identify file types. To extract information from, say, JPEG files, you first have to distinguish them from other files. Reading even the beginning of every file to determine its type would slow indexing considerably.
Windows-specific functions: Qt does not have OS-independent functions to get file system information (volume name, free space, used space, identify a mounted volume) or to eject a disk, so these are done with Windows-specific code. Equivalent code would be needed for each operating system supported.
Case sensitivity of file names: The Windows file system is case-insensitive but case preserving. Other systems are case sensitive. This leads to potential ambiguities when copying comment/category information during re-indexing.
Byte order: Windows systems use little-endian byte ordering. Current code would probably break in many places on a machine using big-endian byte order.

Support for non-English languages

Qt is designed to make it easy to translate the user interface into other languages. Text strings are marked in a way which allows them to be extracted for a (human) translator. A table of translated text is applied at run time. One executable supports multiple languages through different tables. Widgets that make up the user interface resize themselves to make text fit, so if a translated string has a different number of characters than the original, adjustments are mostly automatic. WssIndex follows the standards necessary to allow translation, but no translations currently exist.

Why a GPL release was previously impossible

Here's how you get into a situation where you can't release your source code even if you want to: Compilers typically include source code for library routines. You find a library routine that almost does what you want, but maybe it's buggy or maybe it needs a small change to meet your needs. So you modify the source code and use it. As time goes on, you change compilers multiple times, and the original vendor goes out of business. Now your project depends on code that you're only licensed to distribute as part of a binary. WssIndex for DOS had several such routines; all have been replaced or are no longer needed in WssIndex.

Documentation
If I had time to write good documentation, I'd do something else. [Anon]

WssIndex for DOS had a long manual, which is largely obsolete, in WordPerfect format which is certainly not as popular as it once was. Here are a few important notes:

Program windows save their positions and sizes when closed "normally," i.e. not by using the X at the top right of the window. You'll almost certainly want to adjust sizes and positions the first time you run WssIndex.
The default configuration does not display all available database fields. Use the Table-design option to control which fields are displayed.
For the most part, WssIndex does not use complex queries with multiple criteria. Instead, it uses simple queries with options to add to or select from the currently displayed results.
Watch for bubble help when your mouse hovers over selected dialogs.
It's easy to imagine a disk catalog containing many millions of files. WssIndex keeps the entire catalog in memory, so it will suffer a performance hit if virtual memory needs to be used, and it cannot exceed the 2 GB process space limit imposed by 32-bit Windows. The Qt widget used to display query results takes a significant amount of memory, so it would be a mistake to allow it to use arbitrarily many rows. Instead, query results are broken up into pages of configurable size (default is 1,000 entries 10x or 100x larger is probably ok).
To edit volume names, comments, categories or extra-info, just double-click on the displayed value. To delete a volume, right-click on it in the volume list.
Alternating row colors is recommended to improve readability of layouts that spread file information across multiple lines; elsewhere it's optional.
Qt supports styles, or as most programs call them, skins, to change how the program looks. WssIndex's style menu allows limited skinning. There's a built-in Windows Modern Style, based on code by Michal Mecinski and an option to load a style file. The only useful style file currently available changes the font size on menus and tool bars. Windows Modern Style is only fully implemented under Windows 7 or Vista and is not currently applied when the mouse hovers over the double buttons on toolbars. Styles loaded from files are not currently saved. Style files can be specified on the command line by -stylesheet=sheetname. (Equal sign with no surrounding spaces is required.)
WssIndex is very picky about the tags on mp3 files while players tend to silently ignore problems. To get a "second opinion" on the validity of an mp3 file, try Mp3 Validator or Mp3Utility. These will confirm some, but not all of the errors that WssIndex reports.
Extracting information from self-extracting files such as program installers currently recognizes a rather limited number of formats. Watch for improvements in future releases. Other archive and media file types also need to be added.
Databases written by any previous version of WssIndex can be read; they will be converted to the new format when saved. WssIndex used to put information extracted from files in to the Comment And Category fields. Except for extracted comments, this was probably a bad idea. Such information now has its own field. Existing comment and category data will not be moved.
The sort algorithm used is stable. This means that if, for example, you want to sort with file name as the primary key and file extension as the secondary key, you can click first on the extension column header and then on the file name header and the you'll get the same result as a single sort with two keys. WssIndex can also do a single sort with multiple keys. A single sort is faster, but the user time to set it up will be more than the time saved unless the database is fairly large.

Why did the console-mode version die?

Unicode names: A console mode program can only display 256 different characters. Code pages allow changing which characters can be displayed, but do not increase this limit. Windows uses Unicode for file names which means 64K characters are possible. WssIndex for DOS would sometimes index a file, then be unable to open the file to extract information from it because the conversion from Unicode to Ascii character sets is not reversible. English language users would only occasionally run into character set problems; for other languages it was a real problem with no good solution. (This web page does not use Unicode. As a result, it is unable to properly display the accented characters in Michal Mecinski's name above.)
Long file names: Displaying a table of file names was easy when names were at most 8+3 characters long. Doing so in a text-mode interface with names of essentially unlimited length was awkward and ugly.
Programmer pride: Programming is an art and a programmer should to be proud of his product. As Windows took over the computing world, text-mode programs became anachronisms. A lot of time was wasted on text-mode versions that were never released because even the author didn't like them.

What features have not been ported from the DOS version?

Printing. The problem of nicely displaying potentially long-names on paper is perhaps worse than doing so on screen. CSV (comma separated values) export is available to move results into other programs for custom report generation.
Shell to DOS. Not useful in a multi-tasking operating system.
Keyboard macros. Better implemented by 3rd-party programs.
Support for NDOS. Does anyone use this anymore?
Writing volume labels. Better done at the operating system level.
Import comments. Might be implemented later.
Index-time exclusion lists. These will probably be implemented later.
Interface to anti-virus programs. Anti-virus programs can be configured for on-access scanning without needing any assistance from WssIndex.
Recording of short file names. These aren't really useful. Short names were all the original DOS version supported; later, long names were added while retaining the short names.
Flagging disks as bootable. Recognizing a disk as bootable is an extremely hard problem. WssIndex for DOS assumed any disk containing command.com in the root was bootable. That was an embarrassingly naive algorithm.

Acknowledgments for tools used to create WssIndex

Qt by Nokia. In addition to the build environment and runtime libraries, some of the toolbar icons are borrowed from Qt. Nokia retains copyright for these icons.
MinGW (Minimalist GNU for Windows), the GNU Compiler, library and tools.
Visual Studio Express. Free compilers from Microsoft.
Visual Studio 2010. Non-free compiler from Microsoft, used for the 64-bit version of Wssindex.
Code to add a clear button to text entry fields by girish. (Links in this article are were broken at one point, but now work. The source code is also available in a mailing list archive. Change the file extension from bin to zip to extract.)
Windows Modern Style by Michal Mecinski.
Info-Zip source code.
Nullsoft Scriptable Install System used to create the installer.
Collection of deliberately corrupted archive files used to test robustness of extraction code.
Paint.NET and Irfanview for manipulating icons and animations.
Google, because when you have a programming problem, someone else has probably had the same problem, asked for help, and had an answer posted on the net.