User forums > Help

Code completion does not list enum members

<< < (3/3)

ollydbg:

--- Quote from: alberto on April 22, 2010, 11:31:11 am ---Sample code.....

typedef enum
{
    DYN_NO_DISPLAY = 0,
    DYN_SELECTED,
    DYN_HIGHLIGHTED,
    DYN_NONE,

} T_TR_SCROLL_DYN;


static T_TR_SCROLL_DYN        Tr_Scroller_Dynamic =       DYN_NO_DISPLAY;


Then type Tr_Scroller_Dynamic = ..... no options appear, thought it would provide a choice of the four options above?

Using SVN 6181, on Windows XP SP3.

Please help!

--- End quote ---

Hi, what you want is not implemented in CC. here, when you type:

--- Code: ---Tr_Scroller_Dynamic =|    // your caret are here
--- End code ---

For the currently implementation of CC, we just do a "left" search until we find a non-identifier. For this case, we stop after the "=", and get "nothing" returned.

If you type this:

--- Code: ---Tr_Scroller_Dynamic =DYN_|    // your caret are here
--- End code ---
Then, the "left" search stopped at "=" as well, but now, it returned the "DYN_", so, CC plugin can use "DYN_" as a "search text" to do the suggestion list.

For your feature request, I think it will let the "left" search to parse the whole line, and analyze the whole syntax tree, then if it find the left side of the assignment is a enum, it will list all the enum values in the right side of the "=". I'd like to say, it can be done, but it need some effort and time. :D

nanyu:
I hope CC can do work after we type " = " or "new"

1， =

enum EEE{  E1, E2, E3};
...
EEE e  = |  // <-- here,  cc show a list with items: E1, E2... we can select.

2, new  (VS and C# can do it)

class B {};
class D1 : public B {};
class D2 : public B {};
class C : public D2 {};
...
B* pb = new |  // <-- here, cc show a list with items: B, D1, D2, C ...

sorry for my English.

ollydbg:
@nanyu
This means the "left search" should go beyond the "=", and find the token type before the "=" or "new". We just need to parse the whole statement line instead of partial statement. For example:


--- Code: ---int aaa = Ogre::Root::getSingleton().|

--- End code ---


Till now, the currently implementation just do the analyze below:


--- Code: ---Ogre::Root::getSingleton().|

--- End code ---

The code related in the nativeparser.cpp, in function
BreakUpComponents()..

see the code snippet:

--- Code: ---// Breaks up the phrase for code-completion.
// Suppose the user has invoked code-completion in this piece of code:
//
//   Ogre::Root::getSingleton().|
//
// This function will break this up into an std::queue (FIFO) containing
// the following items (top is first-out):
//
// Ogre             [pttNamespace]
// Root             [pttClass]
// getSingleton     [pttFunction]
// (empty space)    [pttSearchText]
//
// It also classifies each component as a pttClass, pttNamespace, pttFunction, pttSearchText
size_t NativeParser::BreakUpComponents(const wxString& actual, std::queue<ParserComponent>& components)
--- End code ---

So, you can see, we will do a match algorithm one by one, from "Ogre" to "enpty space", and give the suggestion list. A more details can be find in the wiki.

BTW: There are a lot of features (till now, Loaden has released a lot CC patches), and we ( I and Loaden) are discussing about the Tokentree per project instead of Tokentree per workspace. Also, there are a lot of features of CC which we can improve.. We just need more help. :D

nanyu:
@ollydbg:

1, about the "=" operator


--- Quote from: ollydbg on April 25, 2010, 04:59:37 pm ---
--- Code: ---int aaa = Ogre::Root::getSingleton().|

--- End code ---

Till now, the currently implementation just do the analyze below:


--- Code: ---Ogre::Root::getSingleton().|

--- End code ---


--- End quote ---

OH, no. borland C++ builder do this work, but I dislike it, since I't can't do 100% correct thing.

What I want is :

enum EE {e1,e2};
EE e1 =  // <-- here, I type a "space"
int e2 =  //<-- and here..

step1 : I type the "="
step2 : as you said, cc do a "left search". and find e1 is a enum, but e2 is a integer.
          in c++98, enum name is a namesapce...but cc allow we do this:
            EE e1 = EE::  //<--here , cc will show the item list. :)  but I have to delete "EE::" after.
          
step3 : for e1, it show the "enum items list". for e2, do nothing.
         so, the new function will have no bearing on " int aaa = Ogre::Root::getSingleton().| "

------------------
2, about "new" operator

As we known, cc have enough  "class inherited relation" information:

with the code :

--- Code: ---class B {};

class D1 : public B {};

class D2 :  B {};

class E : public D1 {};


--- End code ---

cc got a very good parse result (look the image).  so what I want is :


--- Code: ---step1 :  E = new | << here I press the space key...
step 2 : cc do a parse, and found the "new" ,  and find the "=" ..and find the "E" .. NOW cc can easy to show a list seems like:
E = new |
            | E |
            | D1 |
            | B |

:)


--- End code ---

--------------
I like C::B. and  thinks for you and Loaden do a lot work for c::b.
I hope I can do more work for c::b..
but I'm too busy .... :(.. and my english is too bad...:(((((

[attachment deleted by admin]

Navigation

[0] Message Index

[*] Previous page