Bug 13985 – WebCore/rendering/RenderTable.cpp fails to compile with gcc 4.2.x

• Status: RESOLVED
• Resolution: FIXED
• Component: WebCore Misc.

[Bernhard Rosenkraenzer]

../../../WebCore/rendering/RenderTable.cpp: In member function 'int WebCore::RenderTable::calcBorderLeft() const':
../../../WebCore/rendering/RenderTable.cpp:722: error: no matching function for call to 'max(unsigned int&, const short unsigned int:12&)'
../../../WebCore/rendering/RenderTable.cpp:735: error: no matching function for call to 'max(unsigned int&, const short unsigned int:12&)'
../../../WebCore/rendering/RenderTable.cpp:749: error: no matching function for call to 'max(unsigned int&, const short unsigned int:12&)'
../../../WebCore/rendering/RenderTable.cpp:751: error: no matching function for call to 'max(unsigned int&, const short unsigned int:12&)'
../../../WebCore/rendering/RenderTable.cpp: In member function 'int WebCore::RenderTable::calcBorderRight() const':
../../../WebCore/rendering/RenderTable.cpp:781: error: no matching function for call to 'max(unsigned int&, const short unsigned int:12&)'
../../../WebCore/rendering/RenderTable.cpp:794: error: no matching function for call to 'max(unsigned int&, const short unsigned int:12&)'
../../../WebCore/rendering/RenderTable.cpp:808: error: no matching function for call to 'max(unsigned int&, const short unsigned int:12&)'
../../../WebCore/rendering/RenderTable.cpp:810: error: no matching function for call to 'max(unsigned int&, const short unsigned int:12&)'
make[1]: *** [tmp/RenderTable.o] Error 1

[Bernhard Rosenkraenzer]

Created attachment 14851 [details]
Fix

The problem is that "unsigned int" and "unsigned int:12" are no longer considered the same type.

[David Kilzer (:ddkilzer)]

Comment on attachment 14851 [details]
Fix

Thanks for posting a patch to fix this, Bernhard!

In the future, please set the "review?" flag on patches so that they may be flagged for review, and please add a ChangeLog entry for the updates.  See this page for more details:  http://webkit.org/coding/contributing.html

>-                borderWidth = max(borderWidth, gb.width);
>+                borderWidth = max(borderWidth, (unsigned int)gb.width);

In all of these cases, we prefer to use the static_cast<>() syntax and "unsigned" instead of "unsigned int":

>+                borderWidth = max(borderWidth, static_cast<unsigned>(gb.width));

Please add a ChangeLog entry, update the patch and repost.  Thanks!

[Bernhard Rosenkraenzer]

Created attachment 14859 [details]
Fix adjusted to WebKit coding style

[Bernhard Rosenkraenzer]

Created attachment 14860 [details]
Fix adjusted to WebKit coding style

Forgot to do the unsigned int -> unsigned bit, updated patch

[Sam Weinig]

Comment on attachment 14860 [details]
Fix adjusted to WebKit coding style

This looks fine.

[Mark Rowe (bdash)]

Comment on attachment 14860 [details]
Fix adjusted to WebKit coding style

Your patch is in some weird format that svn-apply does not understand.  Can you please resubmit it in a format that we can actually apply?

[Bernhard Rosenkraenzer]

Created attachment 15140 [details]
Same patch, rediffed to not need patch -p1

The patch was a standard unified diff, you just need to use patch -p1 to apply it because its path names include the name of the top level directory.
Rediffed to apply with patch -p0, which seems to be what svn-apply uses.

[Mark Rowe (bdash)]

Landed in r24058.  Thanks for the patch!

I think the thing that threw svn-apply off was the lack of the "Index: path/to/file" headers in the patch.  Keep in mind that our style guideline requires spaces rather than tabs for indentation, even in the ChangeLog.