• timepp (3/3) Jul 23 2015 I have checked the visual D code, seems the highlighter only do a
• Rainer Schuetze (17/20) Jul 25 2015 You're right, the highlighting just uses lexer information. That's why
• timepp (5/29) Jul 26 2015 I think the following two features would be enough to me:

"timepp" <tongjunhui live.cn> writes:

I have checked the visual D code, seems the highlighter only do a 
very simple lexer parse. is it possible to do the semantic check 
for highlighter (as the tooltip did)?

Rainer Schuetze <r.sagitario gmx.de> writes:

On 24.07.2015 08:55, timepp wrote:
 I have checked the visual D code, seems the highlighter only do a very
 simple lexer parse. is it possible to do the semantic check for
 highlighter (as the tooltip did)?

You're right, the highlighting just uses lexer information. That's why 
identifiers all have the same color (except when explicitely changed, 
see http://rainers.github.io/visuald/visuald/Editor.html).

Visual D uses Alex Bothes semantic engine which is pretty good, but I 
suspect it still produces too many false errors when actually using it 
to show semantic errors (just because D code can be pretty complex to 
analyze).

Just applying it to highlighting might be an intermediate step, as false 
categorization might not be too annoying.

What kind of identifiers are you thinking of highlighting differently?

- types
- templates
- global/tls/local variables
- class/struct/union members
- free functions/member functions/properties

There are so many possible different kinds plus combinations of these...

"timepp" <tongjunhui live.cn> writes:

On Saturday, 25 July 2015 at 07:21:16 UTC, Rainer Schuetze wrote:
 On 24.07.2015 08:55, timepp wrote:
 I have checked the visual D code, seems the highlighter only 
 do a very
 simple lexer parse. is it possible to do the semantic check for
 highlighter (as the tooltip did)?

 You're right, the highlighting just uses lexer information. 
 That's why identifiers all have the same color (except when 
 explicitely changed, see 
 http://rainers.github.io/visuald/visuald/Editor.html).

 Visual D uses Alex Bothes semantic engine which is pretty good, 
 but I suspect it still produces too many false errors when 
 actually using it to show semantic errors (just because D code 
 can be pretty complex to analyze).

 Just applying it to highlighting might be an intermediate step, 
 as false categorization might not be too annoying.

 What kind of identifiers are you thinking of highlighting 
 differently?

I think the following two features would be enough to me:
1. highlight function call/function decl
2. when insertion point inside a variable, highlight all 
occurrences of the same var (like Visual C++)
 - types
 - templates
 - global/tls/local variables
 - class/struct/union members
 - free functions/member functions/properties

 There are so many possible different kinds plus combinations of 
 these...