Code Completion in foreach loop

[koonschi]

Hi,

I like using the foreach statement in c++, provided with the c++11 standard:

Code Select Expand


list<string> strList;
for(string &str : strList)
{
cout << str << endl;
}


But now my problem is, that, in the for block, the variable "str" is unknown by the code completion, which can be quite annoying/confusing.

Is there anything that can be done to resolve this problem?

edit: I'm using the latest nightly build.

Thanks!

Koonschi

[oBFusCATed]

Is there anything that can be done to resolve this problem?

Make a patch that implements this feature.

[killerbot]

this is not purely a problem with the range based for loop, it is a problem with every declaration in the for(; construct, also in declarations in if() construct  [I am not talking about the bodies/blocks of the for/if]

Code Select Expand


if(TxmlElement* problem = handle.FirstChildElement("test"))
{
  /// ---> will not code complete on 'problem'
}


[koonschi]

oBFusCATed: Sounds good. How exactly would that work?

[oBFusCATed]

Here is the code for the latest revision of C::B: http://svn.berlios.de/wsvn/codeblocks/trunk/?rev=8431&peg=8431#a0b40758157c8f16fa703ca3be466fa8a
Go browse it, study it, find where it must be modified, modify it, test the modification, read this http://wiki.codeblocks.org/index.php?title=Creating_a_patch_to_submit_to_BerliOS_%28Patch_Tracker%29 , create a patch, submit to berlios, fix reported bugs.

I hope, I've not missed any of the required steps.  :

[koonschi]

Alright, it seems like it's done .. Is there something like a testing framework or similar where I could add/get some test cases?
I tested multiple cases, seems to be alright though ..
So should I just submit it?

[ollydbg]

Alright, it seems like it's done ..

Good work.

Is there something like a testing framework or similar where I could add/get some test cases?

There is a testing project called "cctest.cbp", its under "trunk\src\plugins\codecompletion", you can build this project, then run the app. (The app will open the file "test.h", and parse it, you can see the log of parserthread or other classes), I think we can use some kind of testing framework like cpp-unit-test in the future.

I tested multiple cases, seems to be alright though ..
So should I just submit it?

Go ahead. Thanks.

[koonschi]

Alright, uploaded it

[ollydbg]

Alright, uploaded it

Good work, I'm testing your patchAdded code completion of declarations in for/if/while heads, Thanks.

[ollydbg]

Some comments:
1, RemoveTemplateArgs is missing in header file, comments needed for this function.
2,

Code Select Expand

   // for loop heads look like this:
   // ([init1 [, init2 ...] ] ; [cond1 [, cond2 ..]], [mod1 [, mod2 ..]])
should have the second ";" instead of ",".

BTW: It is the first time I see you use another instance (smallTokenizer) of Tokenizer class in Parserthread. 

[MortenMacFly]

Good work, I'm testing your patchAdded code completion of declarations in for/if/while heads, Thanks.

Feel free to take over, btw...

[koonschi]

1, RemoveTemplateArgs is missing in header file, comments needed for this function.

I'm confused, I just checked and I thought it was in the patch though? At least if I look at the raw patch file it is all at the bottom ...

Some comments:
2,

Code Select Expand

    // for loop heads look like this:
    // ([init1 [, init2 ...] ] ; [cond1 [, cond2 ..]], [mod1 [, mod2 ..]])
should have the second ";" instead of ",".

Oops, you're absolutely right, I will change that asap.

BTW: It is the first time I see you use another instance (smallTokenizer) of Tokenizer class in Parserthread. 

Well, it is the first time that I contribute to Code::Blocks, too I hope I did not break any rules by using it.
It seemed like the easiest thing to do though, as the expressions in parentheses are parsed as one single token. I did not want to mess around with that, as it's used for function parameters, too.

[koonschi]

Corrected patch is online. RemoveTemplateArgs + comments should be in there, too.

[ollydbg]

Here is my test code:

Code Select Expand


void f()
{
if(TxmlElement* problem = handle.FirstChildElement("test"))
{
  /// ---> will not code complete on 'problem'
     problem;
}

}


void f2()
{
if(TxmlElement* problem1 = handle.FirstChildElement("test"))
{
  /// ---> will not code complete on 'problem'
   problem1;
}
}


And I see the patch works very nice. I right click on "problem", and select find declaration, it goes to the correct position.
When I click on the "problem1", I see the temp Token "problem" is cleared, and new temp Token "problem1" is created.
Nice work, koonschi.

Well, it is the first time that I contribute to Code::Blocks, too Smiley I hope I did not break any rules by using it.
It seemed like the easiest thing to do though, as the expressions in parentheses are parsed as one single token. I did not want to mess around with that, as it's used for function parameters, too.

I hope this is the beginning of your contribution to C::B, welcome!!!!
It dose not break any rules. I have no objection to use another instance of Tokenizer class in Parserthread class.

Besize that, we have a function in Parserthread

Code Select Expand

bool ParserThread::GetBaseArgs(const wxString & args, wxString& baseArgs)
which translate a complex full function arguments to a simple arguments like:
void f(int a=1, float *p=0x777);

here, (int a=1, float *p=0x777) will be translated to (int a, float *p), I think here we also need a smallTokenizer. , because sometimes, there are nested "()" in the function arguments, currently implementation of GetBaseArgs() does not work correctly about those nest parentheses.

[ollydbg]

Bug comes in the function: void ParserThread::HandleConditionalArguments()

Suppose I have a if condition below:

Code Select Expand


if(aaaaa>bbbbb && ccccc<ddddd)

When parsing, it will add a temporary token with the information below:

Code Select Expand


[debug]{m_FullType = "aaaaa&&", m_BaseType = "aaaaa", m_Name = "<", m_Args = "", m_BaseArgs = "", m_AncestorsString = "", m_FileIdx = 1, m_Line = 17, m_ImplFileIdx = 0, m_ImplLine = 0, m_ImplLineStart = 0, m_ImplLineEnd = 0, m_Scope = tsUndefined, m_TokenKind = tkVariable, m_IsOperator = false, m_IsLocal = false, m_IsTemp = true, m_IsConst = 186, m_Index = 501, m_ParentIndex = 500
[debug], m_Children = std::set with 0 elements, m_Ancestors = std::set with 0 elements, m_DirectAncestors = std::set with 0 elements, m_Descendants = std::set with 0 elements, m_Aliases = 0 count of wxArrayString, m_TemplateArgument = "", m_TemplateType = 0 count of wxArrayString, m_TemplateMap = std::map with 0 elements, m_TemplateAlias = "", m_UserData = 0x0, m_TokenTree = 0x3d31270, m_Ticket = 786}>>>>>>cb_gdb:


You see: m_Name is "<".

I think the logic to find a variable:

Code Select Expand


   if (peek.empty())
        {
            if (!m_Str.empty())
            {
             .....
                Token *newToken = DoAddToken(tkVariable, token, smallTokenizer.GetLineNumber());


Here, if peek is empty, and m_Str have some characters, but the token is "logic compare operators".