24 Commits

Author SHA1 Message Date
Vassil Vassilev
14ef88e0ad Avoid double deletion of the ASTContext's external source.
This is another hack, working around the lack of multiplexing external AST source.
2013-10-14 09:11:39 +02:00
Vassil Vassilev
2af379d972 Call InitializeSema on the external sema source. 2013-10-14 09:11:39 +02:00
Vassil Vassilev
fe753e0bb2 Spell it syntactically correct. 2013-09-09 14:33:56 +02:00
Vassil Vassilev
6cb325d52a Try silence an warning. 2013-09-09 14:04:50 +02:00
Vassil Vassilev
1a2a7862c5 Propagate the new callback to our InterpreterCallbacks interface. 2013-09-09 13:14:55 +02:00
Vassil Vassilev
aa5240d5b9 Fix init order warning. 2013-08-22 15:43:55 +02:00
Vassil Vassilev
36b6e36b1a Move the PPCallbacks in InterpreterCallbacks too. 2013-08-22 13:50:57 +02:00
Vassil Vassilev
85179e69e3 Subscribe to the interesting events. 2013-08-22 13:50:56 +02:00
Vassil Vassilev
8c168320f0 Hide the callback adaptor classes in the implementation. 2013-08-22 13:50:56 +02:00
Vassil Vassilev
1c04a16ae6 Hide InterpreterExternalSemaSource class as an implementation detail. 2013-08-22 13:50:54 +02:00
Vassil Vassilev
88cc4a1ed2 Implement autoloading of entities in unknown namespaces. Eg. MyNamespace::MyClass. The implementation is far from ideal, but for the timeline we have it is good enough. On a second iteration it could be improved a lot. 2013-08-21 10:54:51 +02:00
Axel Naumann
70695fe2be Update to clang/llvm r179269 (llvm vendor branch r49215):
Parser::Scope versus Sema::DeclContext are now checked for cross-vailidity.
Emit the TU-transaction explicitly instead of relying on a first transaction.
The typename extraction now takes a stream instead of a string to write to.
The llvm::Linker has much reduced functionality; use llvm::sys::Path instead to find dynamic libraries.


git-svn-id: http://root.cern.ch/svn/root/trunk@49325 27541ba8-7e3a-0410-8455-c3a389f83636
2013-04-24 16:28:08 +00:00
Vassil Vassilev
3bcfc8b25e Prepare dynamic scopes for integration in ROOT;
* Merge the DynamicIDHandler into InterpreterCallbacks. The DynamicIDHandler was
used to define the failed lookup names as dependent, so that later on they could
be "healed" by the EvaluateTSynthesizer - by generating a call to gCling->EvaluateT...

* Add a flag in the InterpreterCallbacks giving a hint when the failed lookup is
performed - during compilation time or during evaluation type, maybe its name is
misnomer. Maybe I should rename it to IsInRuntimeEvalMode or smth like that...

* Simplify the evaluation routine. The idea is to be merged with one of the 
existing - either echo or evaluate

* Implement proper mock object as test dynamic symbol resolver callback. Now it
fully relies on InterpreterCallbacks.

* Implement dynamic expression node marking. Every node that needs runtime 
resolution is marked as dependent. However, clang marks other nodes as dependent,
too (eg. templates). Until now it was very difficult to distinguish between both.
Now every "artificially" (marked by us) node's declaration is annotated so that 
it could be found back easily by using a simple RecursiveASTVisitor.

Generally we have two issues when using dynamic scopes - filtering the failed 
lookups that need to be evaluated at runtime, and filtering the "artificially"
marked as dependent dynamic nodes/decls. Whereas the second issue is solved by
using the annotations, the first one is more tricky to solve because clang doesn't
give us enough information to conclude what should be done. For now I have 
narrowed down the cases so that dynamic nodes can appear only in function decls.


git-svn-id: http://root.cern.ch/svn/root/trunk@48575 27541ba8-7e3a-0410-8455-c3a389f83636
2013-02-14 17:29:30 +00:00
Vassil Vassilev
7832135a63 * Move the check whether a symbol is dynamic from the testsuite to the "mainline".
* Add the dynamic ID handler, responsible for the dynamic scopes, as ExternalSemaSoruce
for the TClingCallbacks. 

From here on we don't know how many tests will fail due to that.
I might end up reverting it if they are too many.


git-svn-id: http://root.cern.ch/svn/root/trunk@48529 27541ba8-7e3a-0410-8455-c3a389f83636
2013-02-11 13:19:55 +00:00
Vassil Vassilev
042f5d6eba Remove extra newline.
git-svn-id: http://root.cern.ch/svn/root/trunk@48528 27541ba8-7e3a-0410-8455-c3a389f83636
2013-02-11 12:56:00 +00:00
Vassil Vassilev
8c2ccd177f Move the DynamicIDHandler into separate source file so that it could be used while
implementing custom interpreter callbacks.


git-svn-id: http://root.cern.ch/svn/root/trunk@48527 27541ba8-7e3a-0410-8455-c3a389f83636
2013-02-11 12:55:01 +00:00
Vassil Vassilev
3b037537f4 The test symbol provider should react on empty lookup result (Thus avoid the ambiguity coming from the extra call put in clang for autoloading).
git-svn-id: http://root.cern.ch/svn/root/trunk@47323 27541ba8-7e3a-0410-8455-c3a389f83636
2012-11-15 16:41:10 +00:00
Vassil Vassilev
20f78c8af1 The external sema source should be set in the constructor of the interpreter
callbacks.
Fixes coverity complaint #47859


git-svn-id: http://root.cern.ch/svn/root/trunk@46531 27541ba8-7e3a-0410-8455-c3a389f83636
2012-10-14 14:20:08 +00:00
Vassil Vassilev
31b81f4212 Silence coverity #47876.
git-svn-id: http://root.cern.ch/svn/root/trunk@46528 27541ba8-7e3a-0410-8455-c3a389f83636
2012-10-14 12:01:32 +00:00
Vassil Vassilev
475ad9a757 * SymbolResolverCallback should create DynamicIDHandler as its InterpreterExternalSemaSource.
* InterpreterExternalSemaSource::setCallbacks is not needed any more. 
  It can be set in the constructor instead.


git-svn-id: http://root.cern.ch/svn/root/trunk@46444 27541ba8-7e3a-0410-8455-c3a389f83636
2012-10-10 15:50:17 +00:00
Vassil Vassilev
9d654b26f9 No need to enable/disable, check for enabled/disabled callback. Either they are
there and we use them or not.


git-svn-id: http://root.cern.ch/svn/root/trunk@46440 27541ba8-7e3a-0410-8455-c3a389f83636
2012-10-10 14:50:45 +00:00
Vassil Vassilev
a047926b87 Usually we have to distinguish between lookup failures. Eg. there are certain
lookups that are expected (by the compiler) to fail and those which should not fail.

I was misled (wasn't thinking at all: Thanks Axel for questioning the checkin) 
because I thought we cannot distinguish those two different cases in the test
callback. 
Silly mistake:
  Revert the changes in the testsuite.
  Revert the default setup of the test callback in enableDynamicLookup

The problem is solved by extracting out the function which determines whether
a lookup is not expected to fail. (It still needs better place.)


git-svn-id: http://root.cern.ch/svn/root/trunk@46438 27541ba8-7e3a-0410-8455-c3a389f83636
2012-10-10 14:37:10 +00:00
Vassil Vassilev
3bba03cf9f *The JIT is configured to compile with RTTI so when we #include the SymbolResolverCallback
it looks for typeinfo of the InterpreterCallbacks (compiled with fno-rtti).

*Move the SymbolResolverCallback into InterpreterCallbacks.(h,cpp). Ideally it is 
used only in the testsuite so we could hack in the build system to build the 
SymbolResolverCallback with fno-rtti and include just the header file but that's
for later.
*Set the symbol resolver callback in enableDynamicLookup for now.
*Clear up the logic behind InterpreterExternalSemaSource and InterpreterCallbacks.
*Now the DynamicIDHandler derives from InterpreterExternalSemaSource.
*Adapt the testsuite to the change.
*Request the InterpreterCallback constructor to be put in the binary.


git-svn-id: http://root.cern.ch/svn/root/trunk@46435 27541ba8-7e3a-0410-8455-c3a389f83636
2012-10-10 13:00:17 +00:00
Vassil Vassilev
e75c9b1a13 The callback class implements LookupObject callback, which should be called even
no dynamic scopes are involved. The information about a failed lookup comes from 
slightly different channel - ExternalSemaSource (if attached to sema).

In order to make the callback LookupObject work when we attach an arbitrary 
callback class we should attach ExternalSemaSource to Sema as well. This require
a default implementation for an ExternalSemaSource.


git-svn-id: http://root.cern.ch/svn/root/trunk@46396 27541ba8-7e3a-0410-8455-c3a389f83636
2012-10-09 10:17:19 +00:00