Bug 18110 – xml document with x-mac-hebrew reports empty document?

• Component: XML
• Status: RESOLVED CONFIGURATION CHANGED

[Eric Seidel (no email)]

xml document with x-mac-hebrew reports empty document?

I'm probably doing something wrong...  See test case.

[Eric Seidel (no email)]

Created attachment 20082 [details]
test case

[Eric Seidel (no email)]

If you remove the <?xml declaration then the document loads just fine.

[Alexey Proskuryakov]

<rdar://problem/3977880>

This is a tricky bug, and we are not aware of any actual sites that use this encoding (not counting test pages as real).

[Alexey Proskuryakov]

I guess some more explanation is required :)

See item 2 and the following item 1 of <ftp://ftp.unicode.org/Public/MAPPINGS/VENDORS/APPLE/HEBREW.TXT> - MacHebrew has '<' as an LTR character, which cannot be mapped to any single Unicode one, which is why decoding to UTF-16 inserts a direction override. The override rightfully confuses the parser, so this is really broken by design.

[Ahmad Saleem]

Both Chrome Canary 119 and Safari 16.6, we get following:

>> error on line 1 at column 44: Unsupported encoding x-mac-hebrew

Do we need anything to do here?

[Anne van Kesteren]

We're not going to support x-mac-hebrew on the web, however, there is a larger question here with what should happen with unknown encodings in XML.

It doesn't appear to be specified in the XML specification. Chromium and WebKit agree on treating them as a fatal error, but Gecko renders the document.

Minimal test:

data:text/xml,<?xml%20version="1.0"%20encoding="a"?><b/>

If at some point we decide to look at XML again maybe it's worth sorting this out.

[Alexey Proskuryakov]

I think that unknown encoding should probably be tracked as a separate bug. This one was very specifically about x-mac-hebrew when it was supported.

What I'm not sure about is whether we have completely removed the support, or if it remains available in some circumstances, such as native apps perhaps.

[Anne van Kesteren]

Renaming back and closing as it's no longer supported per Source/WebCore/PAL/pal/text/TextCodecICU.cpp.