CSV revisited

[Raymond Hettinger]

I think your idea warrants a PEP.

It is the place to collect:
- the pros and cons of the proposal
- what to include/exclude
- various approaches
- References.  For instance, Kernighan & Pike, The Practice of Programming
  devotes a full chapter to issues in building a production CSV library.
  Also, there should be references to code and documentation for CSV
libraries
  in other languages.
- Documentation of key design decisions.  For instance, is the library
limited
   to reading various CSV formats or is write-back in the same style to be
supported.
   Can the first record, contain optional field names?  Will incomplete
records
   raise an exception?
- Pointers to draft implementations so people can experiment with it.

A PEP is an ideal place to collect and document this information.
Also, an addition to the standard library is likely to be considered more
than a 'minor feature request'.  Another advantage is that it will also help
focus the discussion (be prepared, life in the focus is sometimes a little
hot).

If you want a co-author, I'm game.


Raymond Hettinger
"Magnus Lie Hetland" <mlh at vier.idi.ntnu.no> wrote in message
news:slrna62f14.pef.mlh at vier.idi.ntnu.no...
> A little while ago I started a thread about adding CSV support to the
> standard library. There was (as expected) some controversy about this,
> but I think the idea had enough support that it might be turned into a
> formal proposal... However, I'm not sure wheter this is the sort of
> thing that should be formalised as a PEP... Any opinions in the
> matter? (Not pro/con the proposal itself... ;)
>
> --
> Magnus Lie Hetland                                  The Anygui Project
> http://hetland.org                                  http://anygui.org