Rose2xaif

log

age author description
4 weeks ago snarayan Changed include file in generated filedefault tip
4 weeks ago snarayan Changed include file in generated file
4 months ago snarayan Bug fix for annotation generation
4 months ago snarayan Changes for generating ADOL-C files
4 months ago snarayan Changes for generating ADOL-C files
9 months ago utke on macbook this is a hack to make "real(8)" appear in xaif as double FE type
9 months ago utke on macbook relocate up
9 months ago utke on macbook enable "**" operator
9 months ago utke on macbook sweep to remove trailing returns
9 months ago snarayan Fixed the declaration of pointer variables. However this fix is only for single pointers.
9 months ago snarayan Reverted changes on how symbols are looked up
10 months ago snarayan Changed how lookup of symbols for matching of scopes occurs.
10 months ago snarayan Changed how initialized names are created in canonicalization.
10 months ago utke on macbook expanded logic to separate symbols contained in the lists collected by OA but not visible in the current context; see comments on conditions; enhanced debug statements
10 months ago utke on macbook locality determination done here. "isStrictlyLocal" has unclear meaning. The actual question here is whether a symbol is local with respect to a given scope or if a symbol comes from a parent scope. The reference scope for testing here is the body of a function definition. In the search through the parents we
10 months ago utke on macbook settle on one macro
10 months ago utke on macbook dump as c_str
10 months ago utke on macbook dump as c_str
10 months ago utke on macbook don't force it as we had removed the intermediate files from the repo a while ago
10 months ago utke on macbook no such command/group
10 months ago utke on macbook missing parameter doc
10 months ago utke on macbook updated
10 months ago Jean Utke branch merge
10 months ago utke on macbook shouldn't do just formal arguments. need to filter out only strictly locals that aren't formal arguments. everything else should go, particularly globals
10 months ago Jean Utke must decide which language
10 months ago utke on macbook ignore the doxygen output
10 months ago utke on macbook keep this
10 months ago utke on macbook uncomment controlFlowGraphSideEffect call - start fixing up the logic - there is no point in adding the "function" name to the sideeffect list - not sure why this logic was here / rather the original symbol is to be added,,, fix is incomplete as there still is a restriction to loggging only formal symbols which doesn't make much sense by itself...
10 months ago utke on macbook element must be public or else the casts from private element to public graph return 0 - not sure if there is a rule governing this because normally the problem is the other way round that one tries to cast to a private base class
10 months ago utke on macbook formatting only
10 months ago utke on macbook formatting only
10 months ago utke on macbook be able to print elements or graphs
10 months ago utke on macbook make it aware of string values
11 months ago snarayan Enums are handled correctly now
11 months ago snarayan Simplied the canonicalization of SgValueExp nodes, removed dead code as well as commented out junk
11 months ago snarayan Changes to handle per-replacement scopes for reverse mode
14 months ago snarayan C/C++ Command line, ensured that the prefix path is canonicalized
14 months ago snarayan Return if the correct pragma is not found
14 months ago snarayan C/C++ Changed the sparse Saxpy header creation
14 months ago snarayan C/C++ Changed the location of the reversemode template file
16 months ago snarayan C/C++ Backend: conditionals of Preloops are added to the while loop
16 months ago utke on macbook literal constants have a misleading type for Fortran - for now promote them all to double precision - below is an e-mail to Craig to find out if that should be solved in Rose:
16 months ago utke on macbook make the type definition match what happens for constants to be as precise as possible
16 months ago utke on macbook fix up the default type
16 months ago utke on macbook more comprehensive logic for types of constants
16 months ago utke on macbook strip the type
16 months ago utke on macbook set the correct attribute and pass in right away an SgSymbol
16 months ago utke on macbook properly set type and feType for variables and constants
16 months ago utke on macbook print it only if there is something in there
16 months ago utke on macbook capture the output of xaifBooster and convey extra parameters
16 months ago utke on macbook should return a reference, not a copy
16 months ago utke on macbook better error message
16 months ago utke on macbook debug info
16 months ago utke on macbook c/c++ doesn't have the sgsubscripexpression so I separated the logic here and added the bits for handling the bound/stride which then triggers the creation of the SgSubscriptExpression node that isn't there when there is only an index.
16 months ago utke on macbook missing elements in the parser
16 months ago utke on macbook the fix is in the rose main branch now - so we can have the constant folding here
17 months ago snarayan C/c++/Fortran Analysis: The active flag in the SageAttribute is set for symbols that are detected by OA to be active.
17 months ago snarayan Added Activity Analysis support
17 months ago snarayan merge
17 months ago snarayan Changes to pulling of arguments from function calls, handling unary operations and removed some unnecessary debug messages