40601 - RDFServiceImpl::GetUnicodeResource does wrong munging

Status: RESOLVED FIXED

(Core Graveyard :: RDF, defect, P3)

Description

[Viswanath Ramachandran]

From Bugzilla Helper:
User-Agent: Mozilla/4.72 [en] (WinNT; U)
BuildID:    All

RDFServiceImpl::GetUnicodeResource munges non-ascii, non-latin1 unicode uris by 
changing every char > 255 to a '.'. This is problematic. We should instead do 
the munging by converting from Unicode to a UTF8 string. Otherwise many Intl 
uris will all map to the same resource inadvertently. 

Reproducible: Always
Steps to Reproduce:
1. Do GetUnicodeResource for two different Japanese PRUnichar* uris. 
2. They will likely map to the same resource even if they are very different. 


Actual Results:  This means that we cannot use Japanese group names in Instant 
Messenger (unless we do some ugly UTF-8 mangling in our code instead). The right 
place for that to be done is in RDF itself as other people can hit this too in 
the future.

Comment 1

[Chris Waterson]

Need to check in changes for bug 40461 to implement this efficiently.

Comment 2

[Viswanath Ramachandran]

Attachment: Patch for change (NOT COMPLETE)

Comment 3

[Viswanath Ramachandran]

The patch attached before indicates what I would like the behavious of 
GetUnicodeResource to be - however it is NOT CORRECT. Applying this patch means 
the browser doesnt even start and lots of assertions. Looks like the URI munging  
algorithm is used elsewhere in the RDF code or elsewhere ? 

Bugscape 1025 [nsbeta2+] depends on this bug, so I'm nominating it for nsbeta2 
as well. Thanks, Vishy

Comment 4

[Chris Waterson]

Attachment: patch to nsRDFService.cpp; use NS_ConvertUCS2toUTF8

Comment 5

[Chris Waterson]

Fix checked in.

Comment 6

[Chris Waterson]

oops. marking fixed.

Comment 7

[benc]

tever is not RDF QA anymore