Hi, Stanislav, The ibis command always wants a where clause. You can try something like
ibis -d test2 -q "select max(x) where 1=1" On the point about floating-point values, what I wanted to say was floating-point equality is no reliable. The constants used in the query expression in your case come from the string form of the floating-point values, the values in the data table are also from their string representation. In this case, 0.652 matches 0.652. However, if your data were from another source, even if you see 0.652 when you print the values, it may not match the 0.652 you entered from the query expression because the 0.652 might have a lot more (less significant) digits that were not shown to you. Anyway, be careful when you do equality comparison of floating-point values. John On 1/10/13 4:15 PM, Stanislav Seltser wrote: > ok i understand float issues. it makes sense. after i had converted > everything to double the query works. > next thing, why do the sum/max/min/avg functions fail? > ibis -d test2 -q "select max(x)" > > tableSelect:: select(max(x), ) failed on table T-test2 > > > On Jan 10, 2013, at 3:01 PM, [email protected] wrote: > >> Send FastBit-users mailing list submissions to >> [email protected] >> >> To subscribe or unsubscribe via the World Wide Web, visit >> https://hpcrdm.lbl.gov/cgi-bin/mailman/listinfo/fastbit-users >> or, via email, send a message with subject or body 'help' to >> [email protected] >> >> You can reach the person managing the list at >> [email protected] >> >> When replying, please edit your Subject line so it is more specific >> than "Re: Contents of FastBit-users digest..." >> >> >> Today's Topics: >> >> 1. fastbit questions (Stanislav Seltser) >> 2. Re: fastbit questions (Stanislav Seltser) >> 3. Re: fastbit questions (K. John Wu) >> >> >> ---------------------------------------------------------------------- >> >> Message: 1 >> Date: Wed, 9 Jan 2013 23:54:56 -0500 >> From: Stanislav Seltser <[email protected]> >> Subject: [FastBit-users] fastbit questions >> To: [email protected] >> Message-ID: <[email protected]> >> Content-Type: text/plain; charset="us-ascii" >> >> Question 1. ardea parsing >> i have the following small dataset that i store in file data1.csv >> x,y,color >> 0.652,673.114,blue >> 0.255,797.098,green >> 0.541,173.007,green >> 0.231,168.145,red >> 0.233,34.526,red >> 0.219,34.526,red >> 0.722,520.938,blue >> 0.541,438.962,green >> 0.381,797.098,green >> >> i run the following command on it to build index directory >> ardea -b "," -d test2 -b "," -t data1.csv -m x:float,y:float,color:key >> >> >> ardea -d "test2" -v 0 >> Will attempt to parse 1 CSV file >> data1.csv >> with the following column names and types >> x:float,y:float,color:key >> >> >> ardea to read CSV file data1.csv ... >> ardea read 9 rows from data1.csv >> >> now i run ibis query tool >> ibis -d test2 -q "select color" >> >> "blu" >> "gree" >> "re" >> >> what happen to last character?? >> >> Question #2 >> why the exact match query does not work?? >> ibis -d test2 -q "select color where x=0.652" >> >> doQuaere -- "SELECT color WHERE x=0.652" produced a table with 0 row and 0 >> column >> doQuaere -- the result table (0 x 0) for "SELECT color WHERE x=0.652" >> >> but the range query works just fine >> ibis -d test2 -q "select x,color where x <0.66 and x>0.65" >> >> doQuaere -- "SELECT x,color WHERE x <0.66 and x>0.65" produced a table with >> 12 rows and 2 columns >> doQuaere -- the first row (of 12) from the result table for "SELECT x,color >> WHERE x <0.66 and x>0.65" >> 0.652, "blu" >> >> -------------- next part -------------- >> An HTML attachment was scrubbed... >> URL: >> http://hpcrdm.lbl.gov/pipermail/fastbit-users/attachments/20130109/c8708dbf/attachment-0001.htm >> >> >> ------------------------------ >> >> Message: 2 >> Date: Thu, 10 Jan 2013 00:02:02 -0500 >> From: Stanislav Seltser <[email protected]> >> Subject: Re: [FastBit-users] fastbit questions >> To: [email protected] >> Message-ID: <[email protected]> >> Content-Type: text/plain; charset="us-ascii" >> >> so i was to trace the first issue to dos line endings. once i had convert >> the file to unix endings the first question gets resolved. >> the second is still a problem. >> On Jan 9, 2013, at 11:54 PM, Stanislav Seltser <[email protected]> wrote: >> >>> Question 1. ardea parsing >>> i have the following small dataset that i store in file data1.csv >>> x,y,color >>> 0.652,673.114,blue >>> 0.255,797.098,green >>> 0.541,173.007,green >>> 0.231,168.145,red >>> 0.233,34.526,red >>> 0.219,34.526,red >>> 0.722,520.938,blue >>> 0.541,438.962,green >>> 0.381,797.098,green >>> >>> i run the following command on it to build index directory >>> ardea -b "," -d test2 -b "," -t data1.csv -m x:float,y:float,color:key >>> >>> >>> ardea -d "test2" -v 0 >>> Will attempt to parse 1 CSV file >>> data1.csv >>> with the following column names and types >>> x:float,y:float,color:key >>> >>> >>> ardea to read CSV file data1.csv ... >>> ardea read 9 rows from data1.csv >>> >>> now i run ibis query tool >>> ibis -d test2 -q "select color" >>> >>> "blu" >>> "gree" >>> "re" >>> >>> what happen to last character?? >>> >>> Question #2 >>> why the exact match query does not work?? >>> ibis -d test2 -q "select color where x=0.652" >>> >>> doQuaere -- "SELECT color WHERE x=0.652" produced a table with 0 row and 0 >>> column >>> doQuaere -- the result table (0 x 0) for "SELECT color WHERE x=0.652" >>> >>> but the range query works just fine >>> ibis -d test2 -q "select x,color where x <0.66 and x>0.65" >>> >>> doQuaere -- "SELECT x,color WHERE x <0.66 and x>0.65" produced a table with >>> 12 rows and 2 columns >>> doQuaere -- the first row (of 12) from the result table for "SELECT x,color >>> WHERE x <0.66 and x>0.65" >>> 0.652, "blu" >>> >> >> -------------- next part -------------- >> An HTML attachment was scrubbed... >> URL: >> http://hpcrdm.lbl.gov/pipermail/fastbit-users/attachments/20130110/72705015/attachment-0001.htm >> >> >> ------------------------------ >> >> Message: 3 >> Date: Wed, 09 Jan 2013 21:28:55 -0800 >> From: "K. John Wu" <[email protected]> >> Subject: Re: [FastBit-users] fastbit questions >> To: FastBit Users <[email protected]> >> Cc: Stanislav Seltser <[email protected]> >> Message-ID: <[email protected]> >> Content-Type: text/plain; charset=ISO-8859-1 >> >> Hi, Stanislav, >> >> The problem with equality on floating-point number is a known problem >> with the design of the floating-point number scheme. More >> specifically, when you write 0.652 as text, you get three decimal >> digits after the decimal point, however, depending on whether it is >> represented as float or double, you will get different values in the >> computer's memory. In your case, you happen to declare x as float, >> but the value 0.652 in the query expression is represented as a >> double, therefore the in-memory representation of these two values are >> not the same. >> >> In general, one is discourage from perform floating-point value >> equality comparisons in SQL statement - most DBMS systems have the >> issue you are experiencing. If you really need equality comparison, >> convert the number to integer. If you have to keep the values as >> floating-point values, then use a range expression instead. >> >> John >> >> >> On 1/9/13 9:02 PM, Stanislav Seltser wrote: >>> so i was to trace the first issue to dos line endings. once i had >>> convert the file to unix endings the first question gets resolved. >>> the second is still a problem. >>> On Jan 9, 2013, at 11:54 PM, Stanislav Seltser <[email protected] >>> <mailto:[email protected]>> wrote: >>> >>>> Question 1. ardea parsing >>>> i have the following small dataset that i store in file data1.csv >>>> x,y,color >>>> 0.652,673.114,blue >>>> 0.255,797.098,green >>>> 0.541,173.007,green >>>> 0.231,168.145,red >>>> 0.233,34.526,red >>>> 0.219,34.526,red >>>> 0.722,520.938,blue >>>> 0.541,438.962,green >>>> 0.381,797.098,green >>>> >>>> i run the following command on it to build index directory >>>> ardea -b "," -d test2 -b "," -t data1.csv -m x:float,y:float,color:key >>>> >>>> / >>>> / >>>> /ardea -d "test2" -v 0/ >>>> /Will attempt to parse 1 CSV file/ >>>> /data1.csv/ >>>> / with the following column names and types/ >>>> /x:float,y:float,color:key/ >>>> >>>> >>>> ardea to read CSV file data1.csv ... >>>> ardea read 9 rows from data1.csv >>>> >>>> now i run ibis query tool >>>> ibis -d test2 -q "select color" >>>> / >>>> / >>>> /"blu"/ >>>> /"gree"/ >>>> /"re"/ >>>> / >>>> / >>>> /what happen to last character??/ >>>> / >>>> / >>>> /Question #2/ >>>> /why the exact match query does not work??/ >>>> / >>>> ibis -d test2 -q "select color where x=0.652" >>>> >>>> doQuaere -- "SELECT color WHERE x=0.652" produced a table with 0 row >>>> and 0 column >>>> doQuaere -- the result table (0 x 0) for "SELECT color WHERE x=0.652" >>>> / >>>> >>>> but the range query works just fine >>>> ibis -d test2 -q "select x,color where x <0.66 and x>0.65" >>>> >>>> doQuaere -- "SELECT x,color WHERE x <0.66 and x>0.65" produced a >>>> table with 12 rows and 2 columns >>>> doQuaere -- the first row (of 12) from the result table for "SELECT >>>> x,color WHERE x <0.66 and x>0.65" >>>> 0.652, "blu" >>>> >>> >>> >>> >>> _______________________________________________ >>> 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 >> >> >> End of FastBit-users Digest, Vol 65, Issue 4 >> ******************************************** > > _______________________________________________ > 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
