Add nlocal field to DMCompositeLinkRationale: n was used for different purposes, owned part for DM, andlocal part for Array.Also changed DMCompositeGetGlobalISs to return the same number of ISs
Add nlocal field to DMCompositeLinkRationale: n was used for different purposes, owned part for DM, andlocal part for Array.Also changed DMCompositeGetGlobalISs to return the same number of ISs onevery process, instead of only returning for owned part of the Arrayentries. This function was guaranteed to deadlock before because onlyone process was calling ISCreateGeneral(((PetscObject)dm)->comm,...),but this call is necessarily collective.Hg-commit: 4d55b5a7a67069ac539ee945f0eff1496ccfe0f9
show more ...
com->nghost was not set correctly, or is not available yet, so use ltogs to compute the sizeHg-commit: 5ff362d63e435007a1a79407b317275fac95250f
Promote LocalToGlobalMapping and BlockSize from DA to DMHg-commit: adc5020243929ce40bd18f2655d78fe82dc79cc4
Add ISLocalToGlobalMappingGetIndices()Hg-commit: eea7cc64a88fc9a2e3c83f2e385f95d1e1305e39
DMCompositeGather and DMCompositeScatter now accept PETSC_NULL for inactive/unneeded partsHg-commit: d20296ec78b30b2a3702597a5096f2ff597e791c
Small compile fixHg-commit: 78d0c59bc41240f6f432f48399e8bcd7d870e38a
Change ISBlockGetBlockSize to ISGetBlockSize, add ISSetBlockSizeHg-commit: 0a222ea1e9a192436a7e4681121fe012f4928844
Make DMCompositeGather_Array(,ADD_VALUES,) actually aggregate values from all procsHg-commit: 95118fd23e0479b181438de896f590c607404cdb
Implement DMCompositeGetLocalISs()Hg-commit: 7fc6f95569f8d7e3fe7e33229c58a2aa02922409
Change DMCompositeGetGlobalIndices() to DMCompositeGetISLocalToGlobalMappings()Replaces PFIDENTITY and scatter with integer arithmetic. Involves abinary search to locate owning process, because t
Change DMCompositeGetGlobalIndices() to DMCompositeGetISLocalToGlobalMappings()Replaces PFIDENTITY and scatter with integer arithmetic. Involves abinary search to locate owning process, because the VecScatter interalsare not public.Hg-commit: 5fee4c5fd1122c51cc9574943703947fe675e505
DMCompositeGather now takes a single InsertMode parameterHg-commit: bbc80cb501a383f8a663faabcf8ac4e065bffa73
fixed some mansec = DA and mansec = DMDA mistakes from previous changesHg-commit: 5d8cac658961538ec6e588146350f5f4e566d4b7
fix warningHg-commit: a316ed761596ce856b6e10fa51a05e9facc451ed
commit after monster merge with Jed's changes to Truth/Bool broke some stuffHg-commit: 5caec4f276cb70c9322fa9314b3d9d9d41118eac
moved petscda.h and friends over to petscdm.h and friendsHg-commit: 52cf0c5c868ff2e52c7308123a3006fa1da5b856
added DMCreate_DA, DMCreate_ADDA, DMCreate_Sliced, DMCreate_CompositeHg-commit: 69baa2eac668c820d0ef75c58bd3b1a21f98cfbf
finished transistion of DAXXX to DMDAXXXXHg-commit: 86cda14def53e12145f562bc5ffa2be6f2362b57
more conversion of DA methods to DMHg-commit: e2aa391703fc1f000b85cf3ac2e3248e5b4a2830
partial conversion of DA to new DM syntax and styleHg-commit: fc1d8cbf21c8369cb06f934b2976724f2e1219ba
removed DASetType() and all its related infrastructureadded DMSetUp() for time to build data structures from inputted variablesHg-commit: c70bcb5cdb0ec94bd90f78c02694b44ce07f867b
removed DMCompositeGetMatrix() DMSlicedCreateGlobalVector() etc that can simply use DMGetMatrix() DMCreateGlobalVector() etcHg-commit: d661735271f5dc92d48e5b4d58e3b42c9bb2cf3f
moved rest of subclasses of DM into proper subdirectoriesHg-commit: 2dce31395ad72434f195e109111b6e1875e8daa4
moving DM classes examples et into correct directory locationsHg-commit: 43ed329fc2cce22efaa4b439c8485c2b444549fe
1...<<11