Hi, John, Thank you for reply.
I recall another question in my application, and I am studying it further slowly, for I am also a newbie for C++ developer . The situation is that query api crashes when I set more than two thousand “columnA contains aa or” conditions at one where clause. If you have already known the reason about it, I am glad to learn that. Thanks, Roman > On May 17, 2015, at 8:09 AM, K. John Wu <[email protected]> wrote: > > Hi, Roman, > > Thanks for your interest in FastBit. > > Inside FastBit, strings are converted into integers through the > ibis::dictionary class. This class uses a hash function from STL to > do most of the work for converting strings to integers. In terms of > string length, it matters a little bit because every character of a > long string has to be processed in order to generate a hash and every > character has to process in order to compare two strings. However, > the time spent on these operations should be very small and not worth > the trouble for you to convert your long strings into shorter ones. > > Regarding your second question, we will find some time to look into > it, most likely next weekend;-) > > Good luck and feel free to let us know if you have more questions. > > John > > > > On 5/16/15 6:58 AM, Shuangxin Yu wrote: >> >> Hi John, >> >> I am a fresh man in this group . I have two questions when I used >> fastbit to do my work. >> >> 1. To generate fastbit index, I follow two steps. First, ardea command >> to generate fastbit data. Then ibis command to generate fastbit index. >> My question is that whether I can reduce time which costs on >> generating fastbit data by decrease string length for the text type >> column. >> >> 2. I used to generate fastbit index by the version 1.3.8 before. And >> now , I upgrade my fastbit version to 2.0.2, then generating fastbit >> index does not work well for the following parameters（fig1）. It >> seems that the parameter “none” does not work. I compared two version >> of source code about index.cpp, and located the piece of code which >> spec variable compare to “none”,”noindex”. In my opinion the code is >> right, I am confused about the result that building all index for columns. >> >> *$fastbithome/bin/ibis -v -d $indexdir \ >> -b 6 \ >> -b "a:none" \ >> -b "b:none" \ >> -b "c:none" \ >> -b 'd:keywords delimiters=";"' \ >> -b 'e:keywords delimiters=";"' \ >> -b 'f:keywords delimiters=";"' \ >> -b "*:<binning none/><encoding equality/>" -z* >> >> fig1 >> >> Thanks, >> Roman >> >> >> _______________________________________________ >> FastBit-users mailing list >> [email protected] >> https://hpcrdm.lbl.gov/cgi-bin/mailman/listinfo/fastbit-users >> > _______________________________________________ > FastBit-users mailing list > [email protected] > https://hpcrdm.lbl.gov/cgi-bin/mailman/listinfo/fastbit-users _______________________________________________ FastBit-users mailing list [email protected] https://hpcrdm.lbl.gov/cgi-bin/mailman/listinfo/fastbit-users
