Re: [DynInst_API:] Improving dyninst rpm packaging

[Andrew Bernat <bernat@xxxxxxxxxxx>]

As a short answer, DepGraph is Dep-recated (heh) and can be removed. I've fixed most of the others and can push shortly. 

Drew

--
Drew Bernat
bernat@xxxxxxxxxxx
abernat@xxxxxxxxxxx

On Jun 25, 2012, at 10:13 AM, William Cohen <wcohen@xxxxxxxxxx> wrote:

> Hi All,
> 
> There is now a testing rpm based on the dyninst/dyninst_proccontrol branch of the dyninst git repo available in Fedora rawhide:
> 
> http://koji.fedoraproject.org/koji/taskinfo?taskID=4194460
> 
> The package currently builds and installs.  Now working through and seeing what other things need to be fixed.
> 
> The dynint install no longer installs the dynptr.h file from  dynutil/h/dynptr.h.  The body of dynptr.h is ifdef'ed out. There are a number of files that still reference "dynptr.h" in the dyninst/dyninst_proccontrol git branch:
> 
> $ find -type f -exec grep include {} /dev/null \; |grep dynptr.h
> ./depGraphAPI/src/analyzePDG.C:#include "dynptr.h"
> ./depGraphAPI/src/analyzeXPDG.C:#include "dynptr.h"
> ./depGraphAPI/src/DDG.C:#include "dynptr.h"
> ./depGraphAPI/h/xPDG.h:#include "dynptr.h"
> ./depGraphAPI/h/PDG.h:#include "dynptr.h"
> ./depGraphAPI/h/DDG.h:#include "dynptr.h"
> ./depGraphAPI/h/CDG.h:#include "dynptr.h"
> ./depGraphAPI/h/Absloc.h:#include "dynptr.h"
> ./depGraphAPI/h/FDG.h:#include "dynptr.h"
> ./dataflowAPI/src/RoseInsnFactory.h:#include "dynptr.h"
> ./dataflowAPI/h/stackanalysis.h:#include "dynptr.h"
> ./dataflowAPI/h/slicing.h:#include "dynptr.h"
> ./proccontrol/src/response.h:#include "dynptr.h"
> ./proccontrol/h/ProcessSet.h:#include "dynptr.h"
> ./proccontrol/h/Event.h:#include "dynptr.h"
> ./instructionAPI/src/InstructionDecoderImpl.h:#include "dynptr.h"
> ./instructionAPI/src/Operand.C:#include "dynptr.h"
> ./instructionAPI/h/Expression.h:#include "dynptr.h"
> ./instructionAPI/h/InstructionAST.h:#include "dynptr.h"
> ./dyninstAPI/src/ast.h:#include "dynptr.h"
> ./dyninstAPI/src/IAPI_to_AST.h:#include "dynptr.h"
> ./dyninstAPI/src/Relocation/Widgets/Widget.h:#include "dynptr.h" // shared_ptr
> ./dyninstAPI/src/Relocation/Transformers/Transformer.h:#include "dynptr.h" // shared_ptr
> ./dyninstAPI/src/Relocation/CFG/RelocBlock.h:#include "dynptr.h" // shared_ptr
> ./dyninstAPI/src/Relocation/Relocation.h:#include "dynptr.h"
> ./dyninstAPI/src/Relocation/CodeMover.h:#include "dynptr.h"
> ./dyninstAPI/src/function.h:#include "dynptr.h"
> ./dyninstAPI/h/BPatch_snippet.h:#include "dynptr.h"
> ./common/h/singleton_object_pool.h:#include "dynptr.h"
> ./parseAPI/src/IA_powerDetails.h:#include "dynptr.h"
> ./dynutil/h/dynptr.h:#include "boost/shared_ptr.hpp"
> ./dynutil/h/dynptr.h:#include "boost/weak_ptr.hpp"
> ./dynutil/h/dynptr.h:#include "boost/enable_shared_from_this.hpp"
> ./dynutil/h/AST.h:#include "dynptr.h"
> 
> It looks like the following patch is added in the "#include "dynptr.h":
> 
> http://git.dyninst.org/?p=dyninst.git;a=commit;h=2ead336f3d4d504f89fb14f0ed6aad84c0adc96d
> 
> Shouldn't those dynptr.h includes just go away?
> 
> 
> -Will
> _______________________________________________
> Dyninst-api mailing list
> Dyninst-api@xxxxxxxxxxx
> https://lists.cs.wisc.edu/mailman/listinfo/dyninst-api