120357 - AIX's xlC5 compiler can't compile nsBufferHandle.h

Status: RESOLVED FIXED

(Core :: XPCOM, defect)

Description

[Jim Dunn]

dbaron's checkin for bug 114438 broke the xlC 5.x compiler 
while building string/src/nsPromiseFlatString.cpp I get the
following compiler error (very similar to 119326)

"../../dist/include/string/nsBufferHandle.h", line 399.26: 1540-0216 (S) An
expression of type "nsSharedBufferHandle<unsigned short> *" cannot be converted
to "nsSharedBufferHandleWithDestroy<unsigned short> *".

Comment 1

[Jim Dunn]

Attachment: fix that uses a temp variable

Comment 2

[David Baron :dbaron: (⌚️UTC-5, no longer working on Mozilla)]

Seriously, do we really need to support this compiler?  Can't we just use the
old one until they get the bugs out?

Comment 3

[David Baron :dbaron: (⌚️UTC-5, no longer working on Mozilla)]

Also, out of curiosity, does autoconf lead us to use new-style or old-style
casts on this compiler?  If new-style (i.e., static_cast, reinterpret_cast,
const_cast), does the problem go away if we switch it to old-style?

Comment 4

[Jim Dunn]

Attachment: fix to turn off STATIC_CAST on AIX until compiler is fixed

I took dbaron's suggestion and inside of configure.in where we
check for reinterpret_cast (HAVE_CPP_NEW_CASTS) and if this
was aix and we aren't using gcc, then make sure we don't set
this and instead fall back on the old c-style casts (nscore.h)

Comment 5

[Jim Dunn]

Since this is now an configure.in issue, I am hoping to get a
r= from cls so that I can check this in for 0.9.8

Comment 6

[hacker formerly known as [e-mail]]

Comment on attachment 65296 [details] [diff] [review]
fix to turn off STATIC_CAST on AIX until compiler is fixed

r=cls

Comment 7

[David Baron :dbaron: (⌚️UTC-5, no longer working on Mozilla)]

Comment on attachment 65296 [details] [diff] [review]
fix to turn off STATIC_CAST on AIX until compiler is fixed

a=dbaron for 0.9.8

Comment 8

[Jim Dunn]

I removed the keyword.  It turns out that this change causes
all kinds of other problems.  In switching to NS_STATIC_CAST,
we now cast via a (void *) and that causes problems for the
IBM 5.x compiler.  I have submitted the original problem to
IBM for their compiler team to look at.

It is now my opinion that until several issues get resolved
with xlC 5.x compiler, we can't support it (there are others
as well).  This is preventing us from getting a 5.x tinderbox
up so, I don't see how without a tinderbox we can continue to
move forward with this compiler.  I am going to fall back and
stick with xlC 3.6.4.

Comment 9

[Vic Bancroft]

It is not just with the AIX compiler !  This also happens in a crusty,
Linux frege 2.2.14-1mdklinus #1 Tue Jan 4 21:39:25 CET 2000 i686 unknown
gcc version 2.95.2 19991024 (release)


c++ -o nsHTMLTableElement.o -c -DOSTYPE=\"Linux2.2\" -DOSARCH=\"Linux\" -DOJI
-D_IMPL_NS_HTML  -I../../../../dist/include/xpcom
-I../../../../dist/include/string -I../../../../dist/include/gfx
-I../../../../dist/include/layout -I../../../../dist/include/widget
-I../../../../dist/include/dom -I../../../../dist/include/js
-I../../../../dist/include/locale -I../../../../dist/include/unicharutil
-I../../../../dist/include/webshell -I../../../../dist/include/htmlparser
-I../../../../dist/include/necko -I../../../../dist/include/view
-I../../../../dist/include/pref -I../../../../dist/include/docshell
-I../../../../dist/include/xpconnect -I../../../../dist/include/xuldoc
-I../../../../dist/include/caps -I../../../../dist/include/editor
-I../../../../dist/include/gfx2 -I../../../../dist/include/imglib2
-I../../../../dist/include/mimetype -I../../../../dist/include/exthandler
-I../../../../dist/include/uconv -I../../../../dist/include/content
-I../../../../dist/include -I/usr/local/src/mozilla/dist/include/nspr     
-I./../../style/src -I./../../../base/src -I./../../base/src 
-I/usr/X11R6/include   -fPIC  -I/usr/X11R6/include -fno-rtti -fno-exceptions
-Wall -Wconversion -Wpointer-arith -Wbad-function-cast -Wcast-align
-Woverloaded-virtual -Wsynth -Wno-ctor-dtor-privacy -pedantic -Wno-long-long
-pthread -pipe  -DDEBUG -D_DEBUG -DDEBUG_bancroft -DTRACING -g 
-I/usr/X11R6/include -DMOZILLA_CLIENT -include ../../../../config-defs.h
-Wp,-MD,.deps/nsHTMLTableElement.pp nsHTMLTableElement.cpp
In file included from ../../../../dist/include/content/nsRuleNode.h:49,
                 from nsHTMLTableElement.cpp:55:
../../../../dist/include/content/nsCSSDeclaration.h: In method `void
nsCSSDeclaration::AddRef()':
../../../../dist/include/content/nsCSSDeclaration.h:625: warning: comparison is
always true due to limited range of data type
../../../../dist/include/string/nsBufferHandle.h: In method `nsresult
nsHTMLTableElement::SetDir(const nsAString &)':
../../../../dist/include/string/nsBufferHandle.h:399: Internal compiler error.
../../../../dist/include/string/nsBufferHandle.h:399: Please submit a full bug
report.
../../../../dist/include/string/nsBufferHandle.h:399: See
<URL:http://www.gnu.org/software/gcc/faq.html#bugreport> for instructions.
make[5]: *** [nsHTMLTableElement.o] Error 1
make[5]: Leaving directory `/usr/local/src/mozilla/content/html/content/src'
make[4]: *** [libs] Error 2
make[4]: Leaving directory `/usr/local/src/mozilla/content/html/content'
make[3]: *** [libs] Error 2
make[3]: Leaving directory `/usr/local/src/mozilla/content/html'
make[2]: *** [libs] Error 2
make[2]: Leaving directory `/usr/local/src/mozilla/content'
make[1]: *** [tier_9] Error 2
make[1]: Leaving directory `/usr/local/src/mozilla'
make: *** [default] Error 2

Comment 10

[Philip K. Warren]

This bug is fixed with APAR IY27632, which is packaged in the AIX VisualAge C++
Compiler May 2002 update. This also means that the workaround in Bug 119326 is
no longer necessary.

Comment 11

[Jim Dunn]

fixed with xlC 5.0.2.4...
closing