[Psidev-ms-dev] Clarification on mzData specification

[Randy J. <rkj...@in...>]

Based on some recent conversations on the list (and some experience 
helping others develop mzData readers and writers), I suspect we need 
better definition of the base64 arrays.  The last comment was:

> ... The error is similar in each case - for example:
>   Byte data array had length 56 which isn't a multiple of 8
>
The documentation being added to the specification is pretty meager:

spectrumType/mzArrayBinary:
The list of m/z values (for any type of spectrum). The array is stored 
as a base64 encoded binary.The only type allowed is IEEE-754 floating 
point; the precision must be specified as either 32- or 64-bit; 
endianess must also be specified.

The data element (given as a binaryDataGroup) has the attributes:

precision    xs:string    required   
endian       xs:string    required   
length        xs:int        required   

But we don't give a good description of how the data are to be loaded 
into the array.  All of the working implementations are using an 
unpadded binary array which is converted in its entirety to the base64 
string in conformance with: http://www.ietf.org/rfc/rfc3548.txt

In the latest version of the specification document, I have included Kai 
Runte's original base64 utility class implemented in Java as an example 
(Section 4).  Take a look at the developing specification document to 
see how to make the data storage part easier to understand:

http://psidev.sourceforge.net/docstore/view.php?sess=0&parent=7&expand=1&id=35&action=file_details

Thanks,
Randy