76051 - Illegal implicit casts break CodeWarrior6

Status: VERIFIED WONTFIX

(SeaMonkey :: Build Config, defect, P3)

Description

[Andrew Thompson]

CodeWarrior 6 can't build libreg.mcp due to an illegal implicit cast from (char *
*) to (void **). Why do I feel that I'm gonna find a few of these?

Patch will be attached shortly

Comment 1

[Andrew Thompson]

Attachment: Make cast explicit

Comment 2

[Andrew Thompson]

Attachment: Same problem in libpref. (hmm, .c and .cp files being compiled as C++?)

Comment 3

[Andrew Thompson]

Generalised bug, not specific to libreg.

Interesting;y, CW6 seems to be applyng C++ rules to reg.c

Comment 4

[Andrew Thompson]

Attachment: Patch to fix errors building GFX

Comment 5

[Andrew Thompson]

Attachment: Patch to fix casts in plugins

Comment 6

[Andrew Thompson]

Attachment: Fix void ** cast in widget.mcp

Comment 7

[Andrew Thompson]

Attachment: Fixes to InternetConfig and Mime services

Comment 8

[cyrus]

Are we sure there aren't some compiler options that might be a better way to fix
this? How about the "relaxed pointer type rules" option?

Comment 9

[Andrew Thompson]

Um, well, 


* There are only 6 fixes to make in the whole tree.

* Why would we want to disable errors being generated from code which is illegal
according to the C++ spec (or so I'm told).

Obviously one of these errors at least is for a C file and is legal, but having
the casts made explicit actually serves to document what's happening in the code
more clearly, IMO.

Comment 10

[Andrew Thompson]

BTW, the changes I made were purely to make it compile. If anyone can look at
them and say they're conceptually wrong and a better fix it needed, please pipe
up now!

Comment 11

[Simon Fraser [no longer active]]

This one looks suspicious to me:

-       aContext.GetDeviceContext(*getter_AddRefs(theDevContext));
+       aContext.GetDeviceContext((nsIDeviceContext *&)*
getter_AddRefs(theDevContext));

and there's a double cast here which looks wrong:

-  (nsIWidget*)(aEvent->widget)->GetClientData((PluginViewerImpl *)pluginViewer);
+  (nsIWidget*)(aEvent->widget)->GetClientData((void *&)(PluginViewerImpl *
)pluginViewer);

The rest looks good.

Comment 12

[J.J. Enser]

This should not be mine. Reassigning to [e-mail] since he's been doing
all the work

Comment 13

[Steve Dagley]

If this doesn't break CW7 let's close this as WONTIX

Comment 14

[Andrew Thompson]

Well, I don't have 7 to check. Its possible that many of these have been fixed 
along the way in other bugs. Last time I asked somone said all of the CW7 builds 
are on a branch and no fixes have been checked into the trunk. Is this still 
true? If so, we won't know if all of them have been fixed. I don't know if its 
possible to build the trunk in CW 7?

Comment 15

[Steve Dagley]

ccarlen got a branch building on Pro7 and our target to land that on the trunk
is the end of October (2001 for you pessimists :-)  I'm gonna close this and
75333 as WONTFIX.  Sorry to throw away the work Andy but since we skipped Pro6
(and it may very well be that ccarlen used some of this for Pro7 so we're prlooy
not actually throwing it away...

Comment 16

[Jon Granrose]

verified wontfix.

Comment 17

[Jon Granrose]

no, really.