Nate the docs are here (under "documentation" on the main pig site). http://pig.apache.org/docs/r0.8.1/ You are looking specifically for http://pig.apache.org/docs/r0.8.1/udf.html which explains the details of writing loadfuncs .
Bags only have one schema: bags contain tuples. The tuples a bag contains have schemas. Until version 0.9, the only schema maps supported was key:chararray, value:bytearray. Version 9 added support for typed values (keys are still only Strings). I am not sure that there is a way to specify the types for map values from a loader. D On Mon, May 23, 2011 at 12:38 PM, Sweet, Nate <[email protected]> wrote: > Thanks. I guess this means the docs have been wrong for two versions? Uhg. > > Also, ResourceFieldSchema getSchema says only a tuple can have a schema, > which is wrong because it seems a bag can have a schema just fine. Apparently > map cannot have a schema, but I'm just guessing at this point. > > I must say I am very frustrated with the general lack of (and incorrect) > documentation for Pig. I understand the project is evolving rapidly, but IMO > documentation should not suffer. > > -Nate > > > -----Original Message----- > From: Daniel Dai [mailto:[email protected]] > Sent: Friday, May 20, 2011 4:31 PM > To: [email protected] > Cc: Sweet, Nate > Subject: Re: specifying the schema with a LoadFunc > > It is changed to LoadMetadata.getSchema() starting 0.7. > > Daniel > > On 05/20/2011 02:20 PM, Sweet, Nate wrote: >> Hi, >> >> I have a LoadFunc that loads data using a complex schema. I don't want to >> have to specify the schema every time. LoadFunc used to have a method >> "determineSchema". The current docs reference this: >> http://wiki.apache.org/pig/UDFManual#line-595 >> However, the current source does not have this method: >> http://svn.apache.org/viewvc/pig/trunk/src/org/apache/pig/LoadFunc.java?view=markup >> Why was this removed? What replaces this functionality? >> >> Thanks, >> -Nate > >
