People,
what do you think
about language-independent index, in general ?
You see that
current CDT version treats C and C++ as different languages, so for C/C++
mixed project names defined in C files are not visible in C++, vice
versa.
In principle, PDOM
can be language-neutral, but currently it stores C and C++ specific objects -
mostly for handy visual representation in different treeviews. From
navigation's point of view, definition, declaration and reference is
enougth.
So, may be we
could (in future) store some language-independent names to index ? The only
component-specific value can be some integer index related only to graphic
representation (icon), not to
real object
meaning.
Each language (C, C++, C#, Fortran) would provide their
own mapping from their object kinds to corresponding icons.
Other information
(file, position etc) can be the same for every
language.
In this case, we
can have only one (optimized ?) indexing engine for all languages. Specific
language implementations will be responsible for parsing only, in fact.
----------------------------------- With best regards, Oleg
Krasilnikov Software designer, Eclipse team. Intel corp. +7 8312 162
444 ext. 2587 (Russia, Nizhny
Novgorod)
|