Skip to main content

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index] [List Home]
Re: [cdt-dev] Codan does not see symbols from prebuilt index

I will go ahead, open a bug and work on a patch. Thanks.

On Thu, Aug 23, 2012 at 1:01 PM, Sergey Prigogin <eclipse.sprigogin@xxxxxxxxx> wrote:
Hi Eugene,
ADD_EXTENSION_FRAGMENTS_* flags were introduced to allow index extension content providers determine contexts in which the provided extension will be used. We probably need another flag, ADD_EXTENSION_FRAGMENTS_EDITOR, to be used for the shared AST used by the editor.

-sergey

On Thu, Aug 23, 2012 at 11:14 AM, Eugene Ostroukhov <eostroukhov@xxxxxxxxx> wrote:
I traced this to CModelBuilder2::parse - it was:
IIndex index= indexManager.getIndex(fTranslationUnit.getCProject(), IIndexManager.ADD_DEPENDENCIES);
now is:
IIndex index= indexManager.getIndex(fTranslationUnit.getCProject());

Note that ADD_DEPENDENCIES does not revert the behaviour to 8.0.x but I was able to restore it by passing IIndexManager.ADD_EXTENSION_FRAGMENTS_NAVIGATION flag.

What is a proper way to provide symbols that are always visible to particular language?

Best regards,
Eugene


On Thu, Aug 23, 2012 at 10:55 AM, Eugene Ostroukhov <eostroukhov@xxxxxxxxx> wrote:
We have an editor for the custom language that has a standard library that does not require explicitly including the header files.

In CDT 8.0.2 we created a prebuilt index for our library and everything worked wonderfully - code assist, Codan, highlightings, etc. Now we are migrating to 8.1.0 stream and noticed that Codan now highlights those symbols as undefined (code assist, hyperlink navigation still work as before). I am currently debugging it and noticed that Codan uses a different index, the one that does not include the prebuilt index fragment.

Is this a regression or this is a fix for some problem (I understand that old behaviour was not entirely correct - but it is still partially preserved, e.g. in the code assist)? I suppose this change was introduced as a part of allowing including same header file differently.

Should I try to make Codan consider symbols from these fragments or there is another approach to make the editor recognize implicit includes?

Best regards,
Eugene


_______________________________________________
cdt-dev mailing list
cdt-dev@xxxxxxxxxxx
https://dev.eclipse.org/mailman/listinfo/cdt-dev



_______________________________________________
cdt-dev mailing list
cdt-dev@xxxxxxxxxxx
https://dev.eclipse.org/mailman/listinfo/cdt-dev



Back to the top