Bug compatibility

Computer hardware or software is said to be bug compatible if it exactly replicates even an undesirable feature of a previous version. The phrase is found in the Jargon File.[1]

An aspect of maintaining backward compatibility with an older system is that such systems' client programs often do not only depend on their specified interfaces, but also on bugs and unintended behaviour. This must also be preserved by the newer replacement. Besides the significantly higher complexity that needs to be maintained during the natural evolution of the code or interface, this can sometimes cause performance or security issues, and the inconsistencies in the behaviour of interfaces can sometimes lead to new bugs in the software using it, creating difficult to resolve multi-directional cross dependencies between various pieces of code.

Examples of this can be found in MS-DOS / PC DOS, where, when running on 286 or higher processors, the resident executable loader contains code specially designed to detect and fix certain widespread applications and stub loaders (e.g. programs linked with older versions of Microsoft's EXEPACK or Rational Systems' 386 DOS extenders) by patching the loaded program image before executing it, or where DOS patches Windows (WINA20.386)[2] Over the course of development, DR-DOS also had to be modified to not only emulate many undocumented peculiarities of MS-DOS / PC DOS, but also actual bugs in the kernel and several drivers in order to make certain other drivers and applications run on DR-DOS, when they were tested on specific versions of MS-DOS only.[3]

Windows, which has traditionally emulated many old system bugs in order to allow older low-level programs to run, is another example. As a result, Wine, which makes it possible to run many Windows applications on other platforms, also needs to maintain bug compatibility with Windows.[4]

When Microsoft faded out support for 16-bit code in Windows by no longer including NTVDM in 64-bit editions of the operating system, the executable loader was modified to recognize some specific 16-bit stub launchers/installers and replace them on-the-fly with equivalent code stubs running on 64-bit processors.[5][6]

During development of its PC compatible, Compaq engineers found that Microsoft Flight Simulator would not run because of what subLOGIC's Bruce Artwick described as "a bug in one of Intel's chips", forcing them to make their computer bug compatible with the IBM PC.[7] Another hardware example is found in the design of the IBM AT A20 address line to emulate the behaviour in older processors.

Microsoft Excel has always had a deliberate leap year bug which falsely treats February 29, 1900 as an actual date, to ensure backwards compatibility with Lotus 1-2-3.[8]

See also

References

  1. ↑ "Bug-compatible - www.jargon.net". Retrieved 2010-02-03.
  2. ↑ Schulman, Andrew; Brown, Ralf; Maxey, David; Michels, Raymond J.; Kyle, Jim (1994). Undocumented DOS - A programmer's guide to reserved MS-DOS functions and data structures - expanded to include MS-DOS 6, Novell DOS and Windows 3.1 (2 ed.). Addison Wesley. ISBN 0-201-63287-X. ISBN 978-0-201-63287-3.
  3. ↑ Paul, Matthias (2002-02-20). "Need DOS 6.22 (Not OEM)". alt.msdos.programmer. Retrieved 2006-10-14.
  4. ↑ "WineFeatures - The Official Wine Wiki". Retrieved 2010-02-03.
  5. ↑ "Application Installation on 64-bit Systems". Microsoft. Retrieved 2016-05-26.
  6. ↑ "64-bit versions of Windows do not support 16-bit components, 16-bit processes, or 16-bit applications". 2.0. Microsoft. 2011-09-11. KB896458. Archived from the original on 2016-05-26. Retrieved 2016-05-26.
  7. ↑ Yakal, Kathy (January 1985). "Bruce Artwick / The Designer Behind Flight Simulator II". Compute!'s Gazette. p. 32. Retrieved 2014-07-06.
  8. ↑ Excel 2000 incorrectly assumes that the year 1900 is a leap year. Retrieved 2013-09-22.


This article is issued from Wikipedia - version of the 9/7/2016. The text is available under the Creative Commons Attribution/Share Alike but additional terms may apply for the media files.