- Ad-dev - mailman3.cels.anl.gov

[openAD] #202: insufficient parsing of rewind statements
by OpenAD 09 Sep '10

09 Sep '10

#202: insufficient parsing of rewind statements ------------------------------+--------------------------------------------- Reporter: lyonsam | Owner: lyonsam Type: defect | Status: new Priority: minor | Milestone: Component: SourceProcessing | Keywords: ------------------------------+--------------------------------------------- occuring in centrm/pxlib_read_M.f90 (we were able to get around this for now because we dont really need to transform this file): rewind ['(', 'linpxs', ')'] -- Ticket URL: <https://trac.mcs.anl.gov/projects/openAD/ticket/202> OpenAD <http://www.mcs.anl.gov/openad> Bug tracking for OpenAD components

1 4

[openAD] #227: certain character type declarations are not parsed correctly
by OpenAD 09 Sep '10

09 Sep '10

#227: certain character type declarations are not parsed correctly ------------------------------+--------------------------------------------- Reporter: utke | Owner: utke Type: defect | Status: new Priority: major | Milestone: AD for SAS Component: SourceProcessing | Keywords: ------------------------------+--------------------------------------------- example: <pre> CHARACTER(KIND=KIND('A'),LEN=*), OPTIONAL,INTENT(in) :: name </pre> -- Ticket URL: <http://trac.mcs.anl.gov/projects/openAD/ticket/227> OpenAD <http://www.mcs.anl.gov/openad> Bug tracking for OpenAD components

1 4

[ADIC] #47: active type variable declaration (for temporaries) needs to be based on xaif:symbol attribute
by ADIC 08 Sep '10

08 Sep '10

#47: active type variable declaration (for temporaries) needs to be based on xaif:symbol attribute --------------------------+------------------------------------------------- Reporter: utke | Owner: norris Type: defect | Status: new Priority: major | Milestone: Component: XAIF parsing | Version: Keywords: | --------------------------+------------------------------------------------- currently it is based on the xaif:VariableReference attribute which is supposed to express the "usage" (e.g. an activa variable used in a passive context would have active set to false). To declare the variable properly one has to look it up in the xaif:Scope and look at the xaif:Symbol active attribute -- Ticket URL: <http://trac.mcs.anl.gov/projects/ADIC/ticket/47> ADIC <http://trac.mcs.anl.gov/projects/ADIC> ADIC: Automatic Differentiation for C and C++

1 3

[openAD] #225: import of module variables in symbol table fails in _contextReplace
by OpenAD 08 Sep '10

08 Sep '10

#225: import of module variables in symbol table fails in _contextReplace ------------------------------+--------------------------------------------- Reporter: utke | Owner: utke Type: defect | Status: new Priority: major | Milestone: Component: SourceProcessing | Keywords: ------------------------------+--------------------------------------------- The problem is that kind specifications like logical(kind=kind(.true.)) are parsed as _Kind(NamedParam(kind,App(kind,['true']))) while the logic in _contextReplace does and should not expect the expression in _Kind to be a NamedParam. -- Ticket URL: <https://trac.mcs.anl.gov/projects/openAD/ticket/225> OpenAD <http://www.mcs.anl.gov/openad> Bug tracking for OpenAD components

1 2

[ADIC] #46: build errors for [.version] and [install-data-local]
by ADIC 07 Sep '10

07 Sep '10

#46: build errors for [.version] and [install-data-local] ---------------------------------+------------------------------------------ Reporter: utke | Owner: Type: defect | Status: new Priority: minor | Milestone: Component: Configuration/Build | Version: Keywords: | ---------------------------------+------------------------------------------ This a placeholder for fixing the following two errors during the build: gmake[4]: Leaving directory `/sandbox/Argonne/Dev/RoseFE/ADIC_build/runtime' gmake[3]: Leaving directory `/sandbox/Argonne/Dev/RoseFE/ADIC_build/runtime' gmake[3]: [.version] Error 1 (ignored) gmake[3]: Entering directory `/sandbox/Argonne/Dev/RoseFE/ADIC_build' and during the install: cp -pR ./tests/basic/exec/Makefile.incl /sandbox/Argonne/Dev/RoseFE/ADIC_inst/tests/basic/exec mkdir: cannot create directory `/sandbox/Argonne/Dev/RoseFE/ADIC_inst/tests': File exists gmake[4]: [install-data-local] Error 1 (ignored) -- Ticket URL: <http://trac.mcs.anl.gov/projects/ADIC/ticket/46> ADIC <http://trac.mcs.anl.gov/projects/ADIC> ADIC: Automatic Differentiation for C and C++

1 0

[ADIC] #45: faulty expression translation xaif->sage
by ADIC 02 Sep '10

02 Sep '10

#45: faulty expression translation xaif->sage --------------------------+------------------------------------------------- Reporter: utke | Owner: norris Type: defect | Status: new Priority: major | Milestone: Component: XAIF parsing | Version: Keywords: | --------------------------+------------------------------------------------- For the Fortran branch at: /home/utke/HG_Branch/RoseFE/ADIC The backtranslation for a test example yields and incomplete expression for the partial derivative. The test file is attached. -- Ticket URL: <http://trac.mcs.anl.gov/projects/ADIC/ticket/45> ADIC <http://trac.mcs.anl.gov/projects/ADIC> ADIC: Automatic Differentiation for C and C++

1 4

[ADIC] #44: segv during reinterpret cast of SgNode instance address set from string in xaif
by ADIC 04 Aug '10

04 Aug '10

#44: segv during reinterpret cast of SgNode instance address set from string in xaif --------------------------+------------------------------------------------- Reporter: utke | Owner: norris Type: defect | Status: new Priority: major | Milestone: Component: XAIF parsing | Version: Keywords: | --------------------------+------------------------------------------------- This is placeholder to collect info about this problem. Symptom: Program received signal SIGSEGV, Segmentation fault. 0x00002aaab1f7192b in __dynamic_cast () from /usr/lib64/libstdc++.so.6 (gdb) where #0 0x00002aaab1f7192b in __dynamic_cast () from /usr/lib64/libstdc++.so.6 #1 0x00002aaaac2cf691 in isSgStatement(SgNode*) () from /sandbox/Argonne/Apps/rose_inst/lib/librose.so.0 #2 0x00000000005d8169 in adic::SgXAIF::DFSVisitor_XAIF2Sage::onStartElement ( this=0x7fffffffaff0, e=0x2aaae49187f0, empty=true) at ../ADIC/xaif/XAIFVisitors.cpp:296 #3 0x00000000005aee30 in adic::SgXAIF::DFSTraversal::visit ( this=0x7fffffffb9b0, e=0x2aaae49187f0) at ../ADIC/xaif/XAIFGraphTraversals.cpp:119 #4 0x00000000005af0aa in adic::SgXAIF::DFSTraversal::visit ( this=0x7fffffffb9b0, e=0x2aaae4665e60) at ../ADIC/xaif/XAIFGraphTraversals.cpp:136 #5 0x00000000005af4a4 in adic::SgXAIF::DFSTraversal::visit ( this=0x7fffffffb9b0, e=0x2aaae4744050) at ../ADIC/xaif/XAIFGraphTraversals.cpp:157 #6 0x00000000005af0aa in adic::SgXAIF::DFSTraversal::visit ( this=0x7fffffffb9b0, e=0x2aaae408a7e0) at ../ADIC/xaif/XAIFGraphTraversals.cpp:136 #7 0x00000000005af0aa in adic::SgXAIF::DFSTraversal::visit ( this=0x7fffffffb9b0, e=0x2aaae4081d10) at ../ADIC/xaif/XAIFGraphTraversals.cpp:136 #8 0x00000000005ae780 in adic::SgXAIF::DFSTraversal::traverse ( this=0x7fffffffb9b0, e=0x2aaae4081d10) at ../ADIC/xaif/XAIFGraphTraversals.cpp:56 #9 0x000000000060e6b7 in main (myargc=3, myargv=0x7fffffffd3c8) at ../ADIC/main/ADIC.cpp:492 (gdb) p staddr $1 = 18446744073243085568 (gdb) p stmtAddress $2 = "46913461308160" I run on 64 bit opensuse 11.2 with gcc 4.4.1 and have ROSE (pre-release beta version: 0.9.5a) --- using EDG C/C++ front-end version: edg-3.3 --- using OFP Fortran parser version: ofp-0.7.2 and rose svn revision: 779 -- Ticket URL: <http://trac.mcs.anl.gov/projects/ADIC/ticket/44> ADIC <http://trac.mcs.anl.gov/projects/ADIC> ADIC: Automatic Differentiation for C and C++

1 4

[ADIC] #44: Shenzhen Taijie Printing Company in China
by ADIC 26 Jul '10

26 Jul '10

#44: Shenzhen Taijie Printing Company in China -------------------------------+-------------------------------------------- Reporter: Printing in China | Owner: narayanan Type: defect | Status: new Priority: major | Milestone: Component: Canonicalization | Version: Keywords: printing china | -------------------------------+-------------------------------------------- Taijie Printing is a [http://www.china-printing.org/ Printing Company in China] specialized in printing all kinds of printed products, We have more than 10 years experiences in Packaging [http://www.china-printing.org/ Printing in China]; our products including labels, books, catalogs, magazines, calendars, brochures, notebooks, posters, packing materials, advertising collateral, gifts related, CD's and DVD's , etc. Shenzhen Taijie Printing CO. Ltd. Rm 323, Bldg 526, 2nd Bagua Rd, Futian Dist., Shenzhen, 518003 China. Website: [http://www.china-printing.org/ http://www.china-printing.org] -- Ticket URL: <http://trac.mcs.anl.gov/projects/ADIC/ticket/44> ADIC <http://trac.mcs.anl.gov/projects/ADIC> ADIC: Automatic Differentiation for C and C++

1 0

[ADIC] #40: replace debugMsg with a nice macro
by ADIC 23 Jul '10

23 Jul '10

#40: replace debugMsg with a nice macro -------------------------+-------------------------------------------------- Reporter: utke | Owner: Type: enhancement | Status: new Priority: major | Milestone: Component: Utilities | Version: Keywords: | -------------------------+-------------------------------------------------- one of the few things that macros are good for... part of the problem is stuff like this: debugMsg(5, "FILE : " + sg->unparseToString()); The argument "FILE : " + sg->unparseToString() is ALWAYS evaluated regardless of the debug level I specify on the command line. If you replace this with a nice macro like the openad/f one: #define DBG_TAG_MACRO(Group,Tag,StreamArgs) \ if (DbgLoggerManager::instance()->isSelected((Group)) && DbgLoggerManager::instance()->wantTag((Tag))) { \ std::ostringstream aLoNgAnDwEiRdLoCaLnAmeFoRtHiSmAcRoOnLy; \ aLoNgAnDwEiRdLoCaLnAmeFoRtHiSmAcRoOnLy << StreamArgs << std::ends; \ DbgLoggerManager::instance()->logDebug(__FILE__,__LINE__,(Group),aLoNgAnDwEiRdLoCaLnAmeFoRtHiSmAcRoOnLy,(Tag)); \ } you not only avoid evaluating the arguments but also get automatically __FILE__ and __LINE__ injected. Wouldn't that be nice? -- Ticket URL: <http://trac.mcs.anl.gov/projects/ADIC/ticket/40> ADIC <http://trac.mcs.anl.gov/projects/ADIC> ADIC: Automatic Differentiation for C and C++

1 2

[ADIC] #42: adic2 binary not located in standard spot causes odd assumptions for intrinsic file location
by ADIC 23 Jul '10

23 Jul '10

#42: adic2 binary not located in standard spot causes odd assumptions for intrinsic file location -----------------------------+---------------------------------------------- Reporter: utke | Owner: Type: defect | Status: new Priority: minor | Milestone: Component: OA/Sage to XAIF | Version: Keywords: | -----------------------------+---------------------------------------------- e.g. by accident running /ADIC_build/.libs/adic2 gives Fatal Error at file , line 0, char 0 Message: unable to open primary document entity '/sandbox/Argonne/Dev/RoseFE/ADIC_build/share/inlinable_intrinsics.xaif' -- Ticket URL: <http://trac.mcs.anl.gov/projects/ADIC/ticket/42> ADIC <http://trac.mcs.anl.gov/projects/ADIC> ADIC: Automatic Differentiation for C and C++

1 2