Unicode categories breaking change (#20605)

dotnet · Sep 12, 2020 · 3f722a9 · 3f722a9
1 parent ea49f32
commit 3f722a9
Show file tree

Hide file tree

Showing 3 changed files with 66 additions and 0 deletions.
diff --git a/docs/core/compatibility/3.1-5.0.md b/docs/core/compatibility/3.1-5.0.md
@@ -271,9 +271,14 @@ If you're migrating from version 3.1 of .NET Core, ASP.NET Core, or EF Core to v
 
 ## Globalization
 
+- [Unicode category changed for some Latin-1 characters](#unicode-category-changed-for-some-latin-1-characters)
 - [StringInfo and TextElementEnumerator are now UAX29-compliant](#stringinfo-and-textelementenumerator-are-now-uax29-compliant)
 - [Globalization APIs use ICU libraries on Windows](#globalization-apis-use-icu-libraries-on-windows)
 
+[!INCLUDE [unicode-categories-for-latin1-chars](../../../includes/core-changes/globalization/5.0/unicode-categories-for-latin1-chars.md)]
+
+***
+
 [!INCLUDE [uax29-compliant-grapheme-enumeration](../../../includes/core-changes/globalization/5.0/uax29-compliant-grapheme-enumeration.md)]
 
 ***

diff --git a/docs/core/compatibility/globalization.md b/docs/core/compatibility/globalization.md
@@ -9,12 +9,17 @@ The following breaking changes are documented on this page:
 
 | Breaking change | Version introduced |
 | - | :-: |
+| [Unicode category changed for some Latin-1 characters](#unicode-category-changed-for-some-latin-1-characters) | 5.0 |
 | [Globalization APIs use ICU libraries on Windows](#globalization-apis-use-icu-libraries-on-windows) | 5.0 |
 | [StringInfo and TextElementEnumerator are now UAX29-compliant](#stringinfo-and-textelementenumerator-are-now-uax29-compliant) | 5.0 |
 | ["C" locale maps to the invariant locale](#c-locale-maps-to-the-invariant-locale) | 3.0 |
 
 ## .NET 5.0
 
+[!INCLUDE [unicode-categories-for-latin1-chars](../../../includes/core-changes/globalization/5.0/unicode-categories-for-latin1-chars.md)]
+
+***
+
 [!INCLUDE [icu-globalization-api](../../../includes/core-changes/globalization/5.0/icu-globalization-api.md)]
 
 ***

diff --git a/includes/core-changes/globalization/5.0/unicode-categories-for-latin1-chars.md b/includes/core-changes/globalization/5.0/unicode-categories-for-latin1-chars.md
@@ -0,0 +1,56 @@
+### Unicode category changed for some Latin-1 characters
+
+<xref:System.Char> methods now return the correct Unicode category for characters in the Latin-1 range. The category matches that of the Unicode standard.
+
+#### Change description
+
+In previous .NET versions, <xref:System.Char> methods used a fixed list of Unicode categories for characters in the Latin-1 range. However, the Unicode standard has changed the categories of some of these characters since those APIs were implemented, creating a discrepancy. In addition, there was also a discrepancy between <xref:System.Char> and <xref:System.Globalization.CharUnicodeInfo> APIs, which follow the Unicode standard. In .NET 5.0 and later versions, <xref:System.Char> methods use and return the Unicode category that matches the Unicode standard for all characters.
+
+The following table shows the characters whose Unicode categories have changed in .NET 5.0:
+
+| Character    | Unicode category<br>in previous .NET versions | Unicode category<br>in .NET 5.0 and later versions |
+|:------------:|:---------------------------------------------:|:--------------------------------------------------:|
+| § (\u00a7)   | `OtherSymbol`                                 | `OtherPunctuation`                                 |
+| ª (\u00aa)   | `LowercaseLetter`                             | `OtherLetter`                                      |
+| SHY (\u00ad) | `DashPunctuation`                             | `Format`                                           |
+| ¶ (\u00b6)   | `OtherSymbol`                                 | `OtherPunctuation`                                 |
+| º (\u00ba)   | `LowercaseLetter`                             | `OtherLetter`                                      |
+
+#### Version introduced
+
+.NET 5.0 RC1
+
+#### Recommended action
+
+If you have any code that gets the Unicode character category by using the <xref:System.Char> class and assumes the category will never change, you may need to update it.
+
+#### Reason for change
+
+This change was made so that the categories returned by the <xref:System.Char> type are consistent with both the Unicode standard and the <xref:System.Globalization.CharUnicodeInfo> type.
+
+#### Category
+
+- Core .NET libraries
+- Globalization
+
+#### Affected APIs
+
+- <xref:System.Char.GetUnicodeCategory%2A?displayProperty=fullName>
+- <xref:System.Char.IsLetter%2A?displayProperty=fullName>
+- <xref:System.Char.IsPunctuation%2A?displayProperty=fullName>
+- <xref:System.Char.IsSymbol%2A?displayProperty=fullName>
+- <xref:System.Char.IsLower%2A?displayProperty=fullName>
+
+Additionally, any class that depends on <xref:System.Char> to obtain the Unicode character category, for example, <xref:System.Text.RegularExpressions.Regex>, is affected by this change.
+
+<!--
+
+#### Affected APIs
+
+- `Overload:System.Char.GetUnicodeCategory`
+- `Overload:System.Char.IsLetter`
+- `Overload:System.Char.IsPunctuation`
+- `Overload:System.Char.IsSymbol`
+- `Overload:System.Char.IsLower`
+
+-->