#169 Non-ascii characters in string literals

[Anonymous]

This program will not compile with Jikes 1.13. Note the Norwegian letter ř (ISO 8859-1 0xF8), problem is consistent with other higbit characters.

public class JikesBug {
public static void main(String argv[])
{
System.out.println("ř\n");
}
}

Tested on: Redhat Linux 6.2, jdk1.3.
Same problem seems to exist on Solaris 5.7, (jikes 1.12).

jikes-1.11 compiles this program correctly.

Jon.

[Eric Blake]

Bumping priority up.

This example appeared on the jikes mailing list:

if ( count == 1 ) {
pageContext.setAttribute( "id", "1");
pageContext.setAttribute( "description", "Descripci\363n");
}

[Eric Blake]

I consolidated several bugs to use this report. As such, I'm marking the priority up. There may be more than one distinct bug, but they all fit the category of Jikes mis-interpreting 8-bit characters.

[Chris Abbey]

we need to distinguish the bugs that are caused by -encoding misbehaving, and those that are cuased by 8bit input without a -encoding option. None of the bugs you've collected here reference the -encoding option being used, half of them indicate that they have raw 8bit values in their source... this is inherently user error as long as there is not default locale sensitive input (like javac has). In my 1.15 targeted streaming branch these will get a warning due to the ambiguity.

[Chris Abbey]

251 is also related, but it moz let me mark it as dependent. :(

[Chris Abbey]

497 is another dup

[Eric Blake]

Marking this a duplicate of 251, which was fixed in 1.15.