125129 - Status messages not being converted to UTF8

Status: VERIFIED FIXED

(Core Graveyard :: Java: OJI, defect)

Description

[Brian Nesse (gone)]

The MRJ plugin for OS9, and I expect OSX as well, is passing status messages
from the applet to the the plugin support code in the standard Mac character set
rather than converting them to UTF8 which is what the plugin expects to receive.
This is keeping us from displaying status text on any string not in the UTF8
bytecode range.

The applet referenced above should display a copyright notice string in the
status window. This is failing because the copyright symbol "©" has not been
converted to UTF8.

Comment 1

[Brian Nesse (gone)]

Although the code path is completely different on OSX, I see the same results... 
the text does not show up in the status bar. OS -> All

Comment 2

[Brian Nesse (gone)]

Attachment: proposed patch (OS 9 plug-in)

This patch fixes the problem in the OS 9 plug-in.

Comment 3

[Brian Nesse (gone)]

Turns out that the OS X plug-in is trying to do the right thing. But the
destination buffer isn't large enough to handle the additional characters
inserted during the conversion to UTF-8 so it returns an empty buffer instead.

Comment 4

[Brian Nesse (gone)]

Attachment: Proposed patch (OS X)

Patch which fixes the OS X plug-in. Also removes some dead code.

Comment 5

[Steve Dagley]

Comment on attachment 69332 [details] [diff] [review]
Proposed patch (OS X)

r=sdagley (* 2 is probably enough space but what the heck)

Comment 6

[Steve Dagley]

Comment on attachment 69262 [details] [diff] [review]
proposed patch (OS 9 plug-in)

I was going to question the use of delete[] on the result of
::JMTextToEncoding() but I have to question the call itself as I can't find it
on a search of the Apple website and it's not in my JManager.h header.	Am I
missing some update?

Comment 7

[Brian Nesse (gone)]

> (* 2 is probably enough space but what the heck)

To quote the voice from the other side of the wall when I said the same thing
"That is very Euro-centric thinking".

For the normal case, a couple of bytes would probably be fine. However in the
worst case secenario, a Unicode character will be converted into a 3 byte UTF-8
representation. (Technically, according to the doc, you could have up to a 6
byte UTF-8 representation... but this is really special case...)

>Am I missing some update?
JMTextToEncoding is actually a locally defined function:
http://lxr.mozilla.org/seamonkey/source/plugin/oji/MRJ/plugin/Source/MRJContext.cp#847

As to why it has :: in front of it you'd have to ask beard. I stole that code
from something he had written elsewhere. I will remove the :: for clarity.

Comment 8

[Patrick C. Beard]

Comment on attachment 69332 [details] [diff] [review]
Proposed patch (OS X)

sr=beard

Comment 9

[Steve Dagley]

Comment on attachment 69262 [details] [diff] [review]
proposed patch (OS 9 plug-in)

r=sdagley (now that I understand the usage of JMTextToEncoding())

Comment 10

[Patrick C. Beard]

Comment on attachment 69262 [details] [diff] [review]
proposed patch (OS 9 plug-in)

You should make sure |JMTextToEncoding()| isn't returning NULL.

Comment 11

[Brian Nesse (gone)]

Attachment: Better OS9 patch

Add null check for message before displaying and disposing.

Comment 12

[Patrick C. Beard]

Comment on attachment 69690 [details] [diff] [review]
Better OS9 patch

sr=beard

Comment 13

[Steve Dagley]

Comment on attachment 69690 [details] [diff] [review]
Better OS9 patch

r=sdagley

Comment 14

[Patrick C. Beard]

One prefixes calls to functions with :: when one is explicitly calling a 
globally defined function, rather than another member function. It's a C++ 
style thing.

Comment 15

[Brian Nesse (gone)]

Both patches checked in.

Comment 16

[Patty Mac]

verified the patch.