On Dec 1, 2009, at 11:18 PM, Chris Lattner wrote:
>> + /// \brief Specify the point at which code-completion will be
>> performed.
>> + ///
>> + /// \param File the file in which code completion should occur. If
>> + /// this file is included multiple times, code-completion will
>> + /// perform completion the first time it is included. If NULL,
>> this
>> + /// function clears out the code-completion point.
>> + ///
>> + /// \param Line the line at which code completion should occur
>> + /// (1-based).
>> + ///
>> + /// \param Column the column at which code completion should occur
>> + /// (1-based).
>> + ///
>> + /// \returns true if an error occurred, false otherwise.
>> + bool SetCodeCompletionPoint(const FileEntry *File,
>> + unsigned Line, unsigned Column);
>
> This should be moved to libfrontend, I consider this part of the
> 'truncation' half of this.
I just had a "Duh!" moment about this... the preprocessor needs to
know about code completion anyway, because we should be able to
complete things like:
#<list of preprocessor directives>
#if defined(<list of things that are defined>
#pragma <known pragmas>
#define FOO(X, Y, Z)
FOO(blah, <completion containing Y, Z)>
all of which are in the domain of the preprocessor. We just haven't
tried to implement any code completions in the preprocessor yet.
- Doug
_______________________________________________
cfe-commits mailing list
[email protected]
http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits
