Partial updates 6D

[Dimitar Misev]

There is a bug which likely causes an infinite loop (the execution is stuck in the r_Tiler::splitMinterval) when updating a 6D object:

Loaded symbols for /lib/x86_64-linux-gnu/libnss_files.so.2
r_Minterval::operator[] (this=0x7f0f09114700, i=4) at minterval.cc:304
304	    return intervals[i];
(gdb) bt
#0  r_Minterval::operator[] (this=0x7f0f09114700, i=4) at minterval.cc:304
#1  0x00000000006496da in r_Minterval::operator= (this=0x7f0f091146f0, minterval=...) at minterval.cc:340
#2  0x000000000063e37f in std::__copy_move<false, false, std::random_access_iterator_tag>::__copy_m<r_Minterval*, r_Minterval*> (__first=0x7f0f09114700, __last=0x7f0f091c5970, __result=0x7f0f091146f0)
    at /usr/include/c++/4.9/bits/stl_algobase.h:336
#3  0x000000000063de71 in std::__copy_move_a<false, r_Minterval*, r_Minterval*> (__first=0x7f0f09110020, __last=0x7f0f091c5970, __result=0x7f0f09110010) at /usr/include/c++/4.9/bits/stl_algobase.h:396
#4  0x000000000063dfaf in std::__copy_move_a2<false, __gnu_cxx::__normal_iterator<r_Minterval*, std::vector<r_Minterval, std::allocator<r_Minterval> > >, __gnu_cxx::__normal_iterator<r_Minterval*, std::vector<r_Minterval, std::allocator<r_Minterval> > > > (__first=..., __last=..., __result=...) at /usr/include/c++/4.9/bits/stl_algobase.h:434
#5  0x000000000063d654 in std::copy<__gnu_cxx::__normal_iterator<r_Minterval*, std::vector<r_Minterval, std::allocator<r_Minterval> > >, __gnu_cxx::__normal_iterator<r_Minterval*, std::vector<r_Minterval, std::allocator<r_Minterval> > > >
    (__first=..., __last=..., __result=...) at /usr/include/c++/4.9/bits/stl_algobase.h:466
#6  0x000000000063ce78 in std::vector<r_Minterval, std::allocator<r_Minterval> >::_M_erase (this=0x7fff95718f50, __position=...) at /usr/include/c++/4.9/bits/vector.tcc:145
#7  0x000000000063c91d in std::vector<r_Minterval, std::allocator<r_Minterval> >::erase (this=0x7fff95718f50, __position=...) at /usr/include/c++/4.9/bits/stl_vector.h:1150
#8  0x000000000067678e in r_Tiler::splitMinterval (this=0x7fff95718fc0, sourceTile=..., points=std::vector of length 45, capacity 45 = {...}) at tiler.cc:303
#9  0x0000000000675ac4 in r_Tiler::split (this=0x7fff95718fc0) at tiler.cc:82
#10 0x00000000005dd920 in QtUpdate::evaluateTupel (this=0xac7ae10, nextTupel=0x318eab0) at qtupdate.cc:354
#11 0x00000000005dce8c in QtUpdate::evaluate (this=0xac7ae10) at qtupdate.cc:132
#12 0x00000000005f98c7 in QueryTree::evaluateUpdate (this=0xac7b610) at querytree.cc:317

It comes up on

rasql -q 'UPDATE colD6S10Gb AS t set t assign $1' --user rasadmin --passwd rasadmin \
      --mddtype B_MDD_char_6 --mdddomain [0:23,0:23,0:23,23:46,23:46,23:46] \
      --file /e/data/191102976

Tile configuration:

tiling aligned [0:12,0:12,0:12,0:12,0:12,0:12] tile size 4826809

[Dimitar Misev]

Hm I run the query again now on an existing colD6S10Gb and couldn't reproduce the issue. Will try again with recreating the collection from scratch.

[Dimitar Misev]

Can't reproduce it anymore, closing.