Media Type "text/csv" - trailing nullcols

[Yakov Shafranovich]

Tak,

Thanks for your comments. My intent with this draft was to document the 
current convention and register the MIME type. I would rather stay away 
from requiring or allowing features explicetly that might not be 
followed by all implementations.

Yakov

Yakov Shafranovich wrote:
> From: "Tak Tang" <Tak.Tang at suht.swest.nhs.uk>
> To: <ietf-types at alvestrand.no>
> Date: Thu, 24 Mar 2005 11:30:43 +0000
> Subj: Media Type "text/csv" - trailing nullcols
> 
> Hello,
> 
> This is my first post, and I've entered the discussion late, so please 
> excuse me if this has already been discussed . . .
> 
> The draft (03) does not require that each record/line have the same number 
> of values/commas.
> I think it may be benefitial to explicitly allow csv writers to omit 
> trailing commas if there are no further non-null values.
> 
> eg (all of these lines are valid)
> aaa,bbb,ccc,ddd<CR-LF>
> aaa,,,<CR-LF>
> aaa<CR-LF>
> ,bbb<CR-LF>
> ,,ccc<CR-LF>
> ,,,ddd<CR-LF>
> 
> The second and third line are the same - they just have a value in the 
> first field.
> The csv reader should assume that if there are missing columns (ie commas) 
> then the values are null.
> Leading null values still require the commas to be included (lines 4-6).
> 
> This is useful when
> a) the dataset is sparse.
> b) the dataset has new columns added - the reader can still accept old 
> datafiles without them being considered 'malformed'.
> 
> Microsoft Excel and Oracle SQL*Loader can already accept data with missing 
> (trailing) commas.
> 
> 
> Tak Tang
> 
> 
> 
>