Diffing results should not be locale dependent
Posted: Thu Aug 21, 2008 10:19 pm
Hi,
Our Translators and Engineers use CompareIt a lot before checking code in, to make sure that they did not unintentionally corrupt anything.
This is very common especially when we edit multilingual RC files (ANSI encoded). Each language section of the resource file is encoded in its own code page (1252 for western languages, 949 for Korean, 932 for Japanese, etc...).
The problem is that CompareIt seems to assume the system locale when doing its comparison and relies on the OS interpretation of the bytes.
Say I'm on a Korean codepage and I diff a multilingual RC file where Japanese chars were unintentionally corrupted by a Notepad. For example, rev1 of the file has "ƒpƒXƒ
Our Translators and Engineers use CompareIt a lot before checking code in, to make sure that they did not unintentionally corrupt anything.
This is very common especially when we edit multilingual RC files (ANSI encoded). Each language section of the resource file is encoded in its own code page (1252 for western languages, 949 for Korean, 932 for Japanese, etc...).
The problem is that CompareIt seems to assume the system locale when doing its comparison and relies on the OS interpretation of the bytes.
Say I'm on a Korean codepage and I diff a multilingual RC file where Japanese chars were unintentionally corrupted by a Notepad. For example, rev1 of the file has "ƒpƒXƒ