89446 - rdfcat doesn't work with ``http:'' URLs

Status: RESOLVED FIXED

Description

[Chris Waterson]

The ``rdfcat'' test program appears to work correctly with ``file:'' URLs, but
not with ``http:'' URLs. It's possible that it's not driving Necko correctly
anymore. Needs investigation...

Comment 1

[Chase Tingley]

It seems that rdfcat is broken for remote URLs because nsHttpChannel::Open()
is unimplemented -- only AsyncOpen() is supported.  Sincerdfcat doesn't work
with ``http:'' URLs
RDFXMLDataSourceImpl::BlockingParse() returns NS_OK when it can't open the
channel, the failure is silent...

http://lxr.mozilla.org/seamonkey/source/rdf/base/src/nsRDFXMLDataSource.cpp#546
http://lxr.mozilla.org/seamonkey/source/netwerk/protocol/http/src/nsHttpChannel.cpp#1816

I think this broke when the HTTP branch landed in May.

This should probably be sent to Networking, but since it's assigned I'll leave
that for waterson.

Comment 2

[Chase Tingley]

That comment got messed up.  The first paragraph should have read:

It seems that rdfcat is broken for remote URLs because nsHttpChannel::Open()
is unimplemented -- only AsyncOpen() is supported.  Since
RDFXMLDataSourceImpl::BlockingParse() returns NS_OK when it can't open the
channel, the failure is silent...

Comment 3

[Bradley Baetz (:bbaetz)]

darin: What happened to that wrapper class for Open using asyncopen?
threadsafety isn't a problem here.

Comment 4

[Darin Fisher]

i don't want to include a broken non-threadsafe nsHttpChannel::Open
implementation.  however, that being said.. we are pretty close to being able to
implement Open correctly.. there just hasn't been any need (until now).

bbaetz: i still have that wrapper implementation in my mailbox.

Comment 5

[Asa Dotzler [:asa]]

Bugs targeted at mozilla1.0 without the mozilla1.0 keyword moved to mozilla1.0.1 
(you can query for this string to delete spam or retrieve the list of bugs I've 
moved)

Comment 6

[Chase Tingley]

Created attachment 61811 [details] [diff] [review]
fix

Rewritten to use non-blocking Refresh() and pump an event queue until the
datasource loads.  I also corrected the bogus usage and macroized the failure
messages.

The event pumping will generate lots of thread-safety assertions (it looks like
1 per call to ProcessPendingEvents()) for the nsNativeComponentManager, but
since these also occur with the network TestFileTransport app, I wasn't going
to worry about them.

Comment 7

[Chase Tingley]

Created attachment 62362 [details] [diff] [review]
v2

clean diff after dbaron's NS_SetupRegistry changes.

Comment 8

[Chris Waterson]

Comment on attachment 62362 [details] [diff] [review]
v2

sr=waterson

Comment 9

[Chase Tingley]

rjc, can you review?  thanks.

Comment 10

[Robert John Churchill]

Comment on attachment 62362 [details] [diff] [review]
v2

Sure,  r=rjc

Comment 11

[Chase Tingley]

fix checked in.

Comment 12

[benc]

tever is not RDF QA anymore