Fwd: [Quickfix-developers] QuickFix

[Andrei G. <an...@gm...>]

On 6/10/05, Caleb Epstein <cal...@gm...> wrote:
> On 6/10/05, Andrei Goldchleger <an...@gm...> wrote:
> > I am currently evaluating several options for adding FIX support to a
> > large electronic trading system. We need that the solution adheres to
> > the complete FIX specification as much as possible. From my initial
> > evaluation, I believe that QuickFix implements all of the
> > "traditional"(everything but FixML) FIX protocol, including both
> > business messages (described in Vols 3-5)  and the session protocol
> > (Vol 2). Am I right on this one? If I am wrong, can you point any part
>
> Yes.
>
> > of the specification (even if minor) which is not implemented by the
> > current QuickFix version? What about built-in encription, is it
> > supported?
>
> It isn't (see a post from earlier today on this same subject).

OK, so please correct me if I am wrong: in order to implement "built
in encryption", I would have to modify QuickFix to decrypt the
SecureData field according to some pre-defined method, then "add" the
now decrypted fields to a message which also contains fields sent in
plain text, and finally pump the message through validation. I will
also need to negotiate the encryption method during logon, when I
would also setup the encryption method, if applicable.

>
> > Afer reading the documentation, I could not find any reference to
> > FixML support. I know that this is normally considered "extra", but it
> > will be nice  to know if it is supported.
>
> Short answer: It isn't.
>
> Longer answer: There is a method in the QuickFIX Message class to
> "render" it as XML based on an XML Data Dictionary.  However, the tag
> names in FIXML diverge from those in the FIX spec (why, in the name of
> all that is holy did they do that?)

I am not sure if we are talking about the same thing, but if I am not
mistaken some tags were modified in order to save bandwidth. Makes one
wonder why use XML at first place ;-)

> so the output generated using the
> default Data Dictionary files is not FIXML.  I suspect you could whip
> up a Data Dictionary that used the FIXML names and be able to generate
> mostly FIXML-compliant messages.  But I'm sure there will be other
> gotchas (e.g. should attributes be used, or nested text elements) that
> come up.

OK. Can you think of any other specification aspect that is currently
missing from QuickFix?

Again, thanks for all your help,

Andrei Goldchleger