Skip to main content
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index] [List Home]
Re: [cdt-dev] pre-built index fragments and header variants

Hi Mike,

A pre-built index fragment can contain multiple variants of header files just like the project’s PDOM. Multiple variants are triggered by including a header that does not have an include guard with significantly different macro-dictionaries. It does not matter whether the inclusion is made in a source or a header file.

 

A binding in an index fragment is never duplicated, the occurrences (declarations, definitions and references) can be. However, due to the ability to combine multiple fragments we had to deal with such duplications for a long time.

 

It is possible to find all variants of a header (indirectly) included by the file of the editor. This is also true when the headers were picked up from a read-only fragment. However, when performing F3 navigation, we do not use this kind of information for selecting the variant of the header shown in the editor.

 

Markus.

 

From: cdt-dev-bounces@xxxxxxxxxxx [mailto:cdt-dev-bounces@xxxxxxxxxxx] On Behalf Of Mike Kucera
Sent: Thursday, September 13, 2012 11:16 PM
To: CDT General developers list.
Subject: [cdt-dev] pre-built index fragments and header variants

 

Hi guys, Its been a while since I posted here. I have some tough questions for the index experts...

Is it possible to generate a pre-built read-only index fragment that includes multiple variants of certain header files? I would be indexing just a set of header files, so there wouldn't actually be any source files including the headers at the time that the fragment is generated.

If a header is indexed more than once in the same linkage then how are duplicate bindings/names handled? If you did a general text search wouldn't the index potentially return duplicate bindings or names from each variant?

Normally if you navigate (F3) from a source file to a header I'm assuming the index knows in which variant to find the declaration/definition because the include relationship and significant macros are stored in the index. But, if the headers are in a separate read-only fragment then how would it know which variant to use?


Thanks in advance.

Mike Kucera

Rational Developer for Power

Performance Advisor UI Team Lead

IBM Toronto Lab

mkucera@xxxxxxxxxx

 

Back to the top