| Author |
 Topic  |
|
|
.oisyn
Tomato Guru
162 Posts |
 Posted - Nov 26 2007 : 5:57:44 PM
|
First I'd like to suggest to increase the number of characters in the topictitle, because I'm always having difficulties expressing the problem 
The problem is, whenever you use a cv-qualified typedef-id as a template argument or cast destination (that is, whenever it is expressed between angle brackets), VA X decides to color ALL uses of that type in that sourcefile as if it where a variable. This only applies to typedef identifiers, not to the actual user defined types like class names. And also only if you put the cv-qualifiers in front of the type.
struct MyType { };
MyType * ptr = static_cast<MyType*>(0);
const MyType * ptr = static_cast<const MyType*>(0);
typedef MyType MyAlias;
MyAlias * ptr = static_cast<MyAlias*>(0);
const MyAlias * ptr = static_cast<MyAlias const*>(0);
const MyAlias * ptr = static_cast<const MyAlias*>(0); // culprit
The last line makes all occurences of 'MyAlias' be colored like variables. If you remove that last line, it is colored like a type as it should. When using a type directly instead of through a typedef, like in the MyType example, all is fine.
Naturally it also applies when you use volatile or const volatile. |
|
|
feline
Whole Tomato Software
United Kingdom
18942 Posts |
Posted - Nov 27 2007 : 09:09:01 AM
|
I thought I recognised this. I believe this is:
case=2352
turning up in a slightly different context. |
zen is the art of being at one with the two'ness |
 |
|
|
.oisyn
Tomato Guru
162 Posts |
Posted - Jun 03 2008 : 06:47:46 AM
|
This is a friendly reminder that I'm still seeing this in build 1640 
However, it doesn't seem to be limited to typedefs anymore - it now applies to direct types as well, which is pretty annoying as const types between angle brackets (for casting and template parameters) are pretty common. |
|
|
|
feline
Whole Tomato Software
United Kingdom
18942 Posts |
Posted - Jun 03 2008 : 09:14:47 AM
|
If this is getting worse then that is definitely a problem. What sort of code are you seeing a problem with? I have just tried this simple example:
class CTestClassInAngleBrackets { };
static void testGeneralCode()
{
std::vector<const CTestClassInAngleBrackets *> testVec;
}
and for me the colouring is correct. |
zen is the art of being at one with the two'ness |
|
|
|
.oisyn
Tomato Guru
162 Posts |
Posted - Jun 05 2008 : 07:16:54 AM
|
| Move that class definition to a header and you'll notice the bug . Also, it applies to incomplete types in the same sourcefile (if you remove the braces behind the definition of CTestClassInAngleBrackets you'll see the bug appear as well) |
Edited by - .oisyn on Jun 05 2008 07:17:21 AM |
|
|
|
feline
Whole Tomato Software
United Kingdom
18942 Posts |
Posted - Jun 05 2008 : 11:23:09 AM
|
I am seeing the same effect here. Thank you for the clear description. I have put in a new bug for this regression:
case=17359
And I have increased the priority on case=2352 for you which will hopefully help. It is possible the fix for case=17359 will fix both problems, but I really don't know if it will or not. |
zen is the art of being at one with the two'ness |
|
|
|
support
Whole Tomato Software
5566 Posts |
Posted - Jul 14 2008 : 2:52:05 PM
|
case=2352 is fixed in build 1645
case=17359 is fixed in build 1645 |
Whole Tomato Software, Inc. |
|
|
| |
Topic |
|
1619: type coloring when using with cv in < >
