To reproduce: - Visit the URL. - Hit f12 to open the toolbox. - Select the debugger. In the source view area, I see the text: Error loading source: loadSourceError Inspecting the document's <script> element shows the source we ought to be showing.

Likely fallout from bug 905700

In the browser console: Got an exception during SA_onSource: Request failed with status code = 2152398878 in onStopRequest handler for url = http://dahlström.net/svg/html/get-embedded-svg-document-script.html: fetch/streamListener.onStopRequest@resource://gre/modules/commonjs/toolkit/loader.js -> resource://gre/modules/devtools/DevToolsUtils.js:498:29

This bug seems to have existed for a long time, actually. I just tried it on Firefox 32 and it also happens. I just talked to Panos and he said we have a bug handling "punycode", which I've never heard of before but evidently is a real thing for unicode DNS: https://en.wikipedia.org/wiki/Punycode Looks like we pass the URL straight into `Services.io.newChannel`, so not sure if this bug is more at the platform level or not.

D.Script.source.url gives "http://dahlström.net/svg/html/get-embedded-svg-document-script.html", while various non-debugger APIs (for the document, etc.) give the correct URL of. Jim, any idea why D.Script.source.url would mess up string encoding?

(In reply to J. Ryan Stinnett [:jryans] from comment #4) > D.Script.source.url gives > "http://dahlström.net/svg/html/get-embedded-svg-document-script.html", > while various non-debugger APIs (for the document, etc.) give the correct > URL of... "http://dahlström.net/svg/html/get-embedded-svg-document-script.html"

Because DebuggerScript_getUrl does: str = NewStringCopyZ<CanGC>(cx, script->filename()); But NewStringCopyZ expects ISO-8859-1 input or some such nonsense, but script->filename() is UTF-8 (because it needs to be a char* but still represent all of Unicode). Does the JS engine have something better than the LossyUTF8CharsToNewTwoByteCharsZ thing in CharacterEncoding.h to use here?

Bah. Comment 6 is pretty much a word-for-word repeat of bug 987069 comment 1. Can we just fix this? :(