[Psidev-ms-dev] Re: minor change requests

[Chris T. <ch...@eb...>]

Okay I'd like to just answer the points below. The schema will be 
available from Monday next week; CV dev is also proceeding. 
Documentation will follow shortly along with instance files a.s.a.p.


> + The "cvUserParamType" doesn't have explicit provision for storing a 
> double (it has string, int, float, time, URI). It would be very useful 
> for us if it included double. Maybe also the mzArrayBinaryType?

Double is now in there (note though that in the binary data the 
single/double-precision thing is covered by the specification of 
32/64-bit precision for the IEEE-754 'float').


> + "mzArrayBinaryType" really needs an attribute to say what endian the 
> data is stored in.

Now in there too as an attribute.


> + We would like to store information other than doubles in 
> "dataArrayBinary" - e.g. charge state. For 'safety' a type attribute 
> (plus endian) would be useful. We would like options for double, float, 
> int and boolean.

Okay here's something more cogent than I could have managed, that I got 
from someone else. It's pretty unequivocal...

"I'm a bit opposed to introducing other number formats to the base64 
encoded arrays. In principle integers and even boolean can be converted 
into IEEE floating points (although it seems, at least for boolean, 
terribly wasteful). And if we introduce other formats to the base64 
encoded arrays, we have to _very_ specific how these formats are encoded 
(e.g. how do Java, C++, C# and Perl handle integer or boolean internally 
-- I don't believe it's in the same way). To be absolutely sure we would 
need to provide implementations for these encodings for all sorts of 
programming languages and even we could (and would) run into problems as 
certain labs/vendors/etc. wouldn't like our implementation (or simply 
wouldn't know about) and come with their own (incompatible) 
implementation..."


> + It's probably been discussed before, but what will instrument vendors 
> do when they don't have the required information:
>    personType:  name, institution

Hopefully something meaningful (dept. head + company?); but if all else 
fails I believe there's a Dr Null Foobar, at the Cipher Institute whose 
name we can abuse ;)


Cheers, Chris.

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
  Chris Taylor (ch...@eb...)
  HUPO PSI: GPS -- psidev.sf.net
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~