143461 - nsWebBrowserPersist.cpp creates internal compiler error in xlC 3.6.4

Status: VERIFIED FIXED

Description

[David Favor]

The line:

   msgId = aIsReadError ? NS_LITERAL_STRING("readError") :
                          NS_LITERAL_STRING("writeError");

creates an internal compiler error. I have tried several ways to rewrite
this line and am unsure if I am doing it correctly. Let me know how you
would rewrite this line.

Thanks.

Comment 1

[Jim Dunn]

if (aIsReadError) {
    msgId = NS_LITERAL_STRING("readError");
} else {
  msgId = NS_LITERAL_STRING("writeError");
}

Comment 2

[Adam Lock]

What is xlC?

The if/else should do the trick, but it sounds like the compiler has a problem
(with Unichar strings?) if it can't compile something like this.

Comment 3

[Jim Dunn]

xlC is the c++ compiler on AIX.
Yes xlC has a problem but it is based on how
   msgId = aIsReadError ? NS_LITERAL_STRING("readError") :
                          NS_LITERAL_STRING("writeError");

and how temporary variables/registers are used and then 
assigned.

The fact that msgID = NS_LITERAL_STRING("readError")
works but the above doesn't indicates that it is not so much
an issue with Unichar strings but more on the code syntax
and how NS_LITERAL_STRING is defined.

Comment 4

[Adam Lock]

CC some string gurus

Comment 5

[Alec Flett]

does the if() version work? if so, we should just go with that. what type is msgId?

Comment 6

[Adam Lock]

msgId is an nsAutoString:

http://lxr.mozilla.org/seamonkey/source/embedding/components/webbrowserpersist/src/nsWebBrowserPersist.cpp#873

Comment 7

[Jim Dunn]

Created attachment 84311 [details] [diff] [review]
This patch works for me on xlC 3.6.4

Comment 8

[Kathleen :Brade]

Comment on attachment 84311 [details] [diff] [review]
This patch works for me on xlC 3.6.4

r=brade

Comment 9

[Alec Flett]

Comment on attachment 84311 [details] [diff] [review]
This patch works for me on xlC 3.6.4

sr=alecf

Comment 10

[Jim Dunn]

fix checked into trunk

Comment 11

[benc]

reporter: can you verify the fix?

Comment 12

[David Favor]

I am no longer compiling mozilla, so am unable to verify the fix.

Comment 13

[David Favor]

You bet.