OpenADFortTk

log

find changesets by author, revision, files, or words in the commit message
rev 1210: (0) -1000 -300 -100 -60 tip
age author description
13 days ago utke on macbook SCAN has 2 string arguments and needs the implicit parameters set, but for other string intrinsics there is some special logic somewhere that can't handle the implicit parameters so I do it for SCAN and nothing elsedefault tip
13 days ago utke on macbook type for SCAN
2 weeks ago utke on macbook added scan intrinsic
3 months ago Jean Utke merge
7 months ago azamat Adding handling of non-array pointers.
7 months ago utke on macbook also look for PSTORE
7 months ago utke on macbook pointer assignment interface for DUGactivity
4 months ago utke on macbook handling non-hoisted operations
4 months ago utke on macbook debug message
9 months ago hlcm fixed handling of external declaration for function pointers; added "void" shape and "pointer" kind
10 months ago utke on macbook fix up references
10 months ago utke on macbook re-autogen after merge
10 months ago utke on macbook branch merge
10 months ago utke avoid name clashes on case insensitive file systems
11 months ago utke on macbook add ichar
11 months ago utke on macbook add shape/reshape
12 months ago hcm correct descent order in ConvertScalarizedRefToActiveType
12 months ago hcm add minloc intrinsic
13 months ago utke on macbook get the answer right for strctfld
13 months ago utke on macbook proper conversion for scalars (using field id instead of offset) and plain array types
13 months ago utke on macbook symbol type derives from RefObjType which will drill down but the shape must come from the FLD tree type, i.e. when we are at the ILOAD we need to go down one step.
13 months ago utke on macbook fld is accessed by id, not by offset; need to have logic to deal with type conversion for array members.
13 months ago utke on macbook the scalarized ref may be associated with the ILOAD or the STRCTFLD - have to cover both cases
13 months ago utke on macbook the scalarized ref may be associated with the ILOAD or the STRCTFLD - have to cover both cases
13 months ago utke on macbook there are other constructs that are not pointers and therefore shouldn't go in the else branch where they are assumed to be pointers
13 months ago utke on macbook for debugging
13 months ago utke turn this into a warning
13 months ago utke remove things that are preprocessed away
13 months ago utke determine the actual shape, don't hardwire it to scalar
13 months ago utke STRCTFLD doesn't have an ST so the type has to be determined otherwise
13 months ago utke formatting only
13 months ago utke separate OPR_STRCTFLD from the rest and remove the check forcing the final type to scalar
14 months ago utke progress output
14 months ago utke clarification about maxval/minval
14 months ago utke when the operand is a strctfld
14 months ago utke all the logic is in xlate_MemRef
14 months ago utke if it is a strctfld we need to test agains it, not the ref_wn
14 months ago utke needed to restore the scalarized ref
14 months ago utke the exprTree doesn't has an empty string output - switch to the DumpWN string to get a decent and distinguishable hash
14 months ago utke need to use the C-string
14 months ago utke provide string rep
14 months ago utke make the resetting conditional on the refWN existence
14 months ago utke The alias map key etc are looked up based on the most recent WN in the context stack. if we don't reset the most recent WN here, it stays potentially at the level of the ISTORE etc, i.e. we get instead of the alias set ID for the parameter Memref, the alias set id for the LHS of the whole assignment. This then leads to incorrect assumptions about data dependencies between statements as reported in http://trac.mcs.anl.gov/projects/openAD/ticket/271.
14 months ago utke debug info
14 months ago utke suppress selectors for size arguments
14 months ago utke avoid conversions for SIZE
14 months ago utke cover maxloc
14 months ago utke cover maxloc
14 months ago utke suppress conversions
14 months ago utke nullify
16 months ago utke changed things to make doxygen happy
16 months ago utke try this
16 months ago utke obsolete
16 months ago utke use canonical names. reflected in intrinsic catalogue
16 months ago utke implement extra debug interface
17 months ago utke prevent misleading array bounds to show up for assumed size arrays
17 months ago Jean Utke branch merge
23 months ago utke 64 bit mac os
18 months ago utke unused
19 months ago utke cover _(DE)ALLOCATE
rev 1210: (0) -1000 -300 -100 -60 tip