121833 - ASSERTION: |First()| on an empty string: 'Length()>0'

Status: RESOLVED FIXED

(Core Graveyard :: RDF, defect, P3)

Description

[timeless]

#3  0x282fb76c in nsAString::First (this=0xbfbfa428) at /home/timeless/mozilla/string/src/nsAString.cpp:140
#4  0x28d1a969 in rdf_MakeRelativeRef (aBaseURI=@0xbfbfa394, aURI=@0xbfbfa428)
    at /home/timeless/mozilla/rdf/base/src/rdfutil.cpp:84
#5  0x28d17af9 in nsRDFXMLSerializer::SerializeDescription (this=0x8bcd0c0, aStream=0x8b89604,
    aResource=0x8bf7360) at /home/timeless/mozilla/rdf/base/src/nsRDFXMLSerializer.cpp:535
#6  0x28d1a7ca in nsRDFXMLSerializer::Serialize (this=0x8bcd0c0, aStream=0x8b89604)
    at /home/timeless/mozilla/rdf/base/src/nsRDFXMLSerializer.cpp:1040
#7  0x28d13db6 in RDFXMLDataSourceImpl::Serialize (this=0x8b9c300, aStream=0x8b89604)
    at /home/timeless/mozilla/rdf/base/src/nsRDFXMLDataSource.cpp:1096
#8  0x28d130b9 in RDFXMLDataSourceImpl::Flush (this=0x8b9c300)
    at /home/timeless/mozilla/rdf/base/src/nsRDFXMLDataSource.cpp:831
#9  0x282ea9b8 in XPTC_InvokeByIndex (that=0x8b9c304, methodIndex=6, paramCount=0, params=0xbfbfaabc)
    at /home/timeless/mozilla/xpcom/reflect/xptcall/src/md/unix/xptcinvoke_unixish_x86.cpp:153

Comment 1

[timeless]

Attachment: try to be nice

Comment 2

[Chris Waterson]

Attachment: wouldn't this be simpler?

Don't we just need to check if aURI IsEmpty?

Comment 3

[Chase Tingley]

Comment on attachment 66459 [details] [diff] [review]
wouldn't this be simpler?

r=tingley.

rdf_MakeRelativeName looks like it could have the same issue, but I'm not sure
it would ever actually arise.  (Actually, I'm not sure the function is ever
called.)
A couple other spots with this pattern too (one in nsRDFXMLSerializer.cpp that
should be fine, and one in nsRDFParserUtils::StripAndConvert
that looks like it might be possible to trigger).

Comment 4

[Chris Waterson]

Attachment: remove unused function, too

Sho'nuff. rdf_MakeRelativeName is unused. Removed it. I think the usage of
`First' in nsRDFXMLSerializer.cpp should be fine, and the use in
nsRDFParserUtils.cpp is already protected by an IsEmpty check.

Comment 5

[Mike Shaver (:shaver emeritus)]

Comment on attachment 66470 [details] [diff] [review]
remove unused function, too

sr=shaver.

Comment 6

[Chris Waterson]

Attachment: one more time

I found timeless on IRC and it turns out he was trying to optimize this code,
as well. The string-fu here is archaic: this last patch cleans up timeless's
optimization, and uses Substring rather than Find.

Comment 7

[timeless]

Comment on attachment 66485 [details] [diff] [review]
one more time

i like it.

Comment 8

[Mike Shaver (:shaver emeritus)]

Comment on attachment 66485 [details] [diff] [review]
one more time

Yeah, sure, whatever. sr=shaver.

Comment 9

[timeless]

Comment on attachment 66485 [details] [diff] [review]
one more time

r=timeless [patch manager collision]

Comment 10

[Chris Waterson]

Fixed. Thanks!