It is a bug. I have created a jira and attached a patch -
https://issues.apache.org/jira/browse/PIG-2223
-Thejas
On 8/16/11 3:16 PM, [email protected] wrote:
Hello,
I'm trying to generate a tuple from a very wide data set, but running in
to problems. I'm running Pig 0.9.0 r1148983 in local mode.
Because the data set it so wide, I'd prefer not to explicitly state each
field, and instead use either * or the $0..$N syntax when generating the
tuple. Additionally, the schema of each field is quite long and
arbitrary as the data has been generated using macros. (Which is
essentially why I'm putting the data into a tuple - so that further down
in my script I can more easily refer to the fields)
The code below illustrates the issue I'm having referencing fields in
the tuple. Is it a bug, or am I missing something?
Thanks in advance,
Grahame
describe a;
a: {f1: int,f2: int,f3: int,f4: int,f5: int,f6: int,f7: int,f8: int,f9:
int,f10: int}
aa = FOREACH a GENERATE $0, TOTUPLE($2,$3,$4,$5);
aaa = FOREACH aa GENERATE $0, $1.$0; -- OK
aaa = FOREACH aa GENERATE $0, $1.f3; -- OK
aaa = FOREACH aa GENERATE $0, $1.$1; -- OK
aaa = FOREACH aa GENERATE $0, $1.f4; -- OK
aa = FOREACH a GENERATE $0, TOTUPLE($2..$5); -- should be the same as above?
aaa = FOREACH aa GENERATE $0, $1.$0; -- OK
aaa = FOREACH aa GENERATE $0, $1.f3; -- ERROR
aaa = FOREACH aa GENERATE $0, $1.$1; -- ERROR
aaa = FOREACH aa GENERATE $0, $1.f4; -- ERROR
aa = FOREACH a GENERATE $0, TOTUPLE(*);
aaa = FOREACH aa GENERATE $0, $1.$0; -- ok
aaa = FOREACH aa GENERATE $0, $1.f3; -- ERROR
aaa = FOREACH aa GENERATE $0, $1.$1; -- ERROR
aaa = FOREACH aa GENERATE $0, $1.f4; -- ERROR
