Wikipedia:Niqqud

From Wikipedia, the free encyclopedia

For the past several months there has been a serious technical problem with the Unicode representation of Hebrew vowel signs (niqqud) on Microsoft Windows XP, affecting users editing Wikimedia projects when running this operating system.

Upgrading to Windows XP SP 2 fixes this bug.

The problem is especially severe because it not only affects how the vowels are seen on the screen, but apparently changes the actual Unicode coding of the Hebrew letters and vowels when submitted via the browser form. I.e. after a text that has been correct in the past coded undergoes "save" or even "show preview" - the coding is altered in a way that makes it nearly impossible to read.

Note that correctly coded texts from before the recent Wikimedia upgrade (preserved in the "History" of the page) retain the proper coding. But any attempt to revert to a version with the correct coding is automatically destroyed.

The problem has been noted several times at the Hebrew Wikipedia over the past few months, and a serious discussion has begun here. This parallel English page has been created so that the discussion will be available in English, hopefully facilitating a solution to the bug.

[edit] Example

The following link gives a very clear, typical example of the problem:

• Minor change to vowelized Hebrew text by adding category

What happened at that link is the following:

1. A vowelized text (i.e. one with niqqud) was added to Hebrew Wikisource on December 9, 2004.
2. A well-meaning user made minor edit to that text (he simply added a category tag at the end) on June 7, 2005. But when that user pressed "save" it created a new version of the entire text, as can be seen from the comparison in the link: The new version contains vowel changes in the majority of the text's words! This happened automatically, and is extraordinarily difficult to correct manually (even if the software allowed fixing it).

[edit] Current status links

The following page contains an up-to-date summary of the problem by the developers, and is the best place to present new developments or ask questions (in addition to the bug page):

• mw:Unicode normalization considerations
• bugzilla:2399 (with response by Ken Whistler, Unicode 5.0 editor)