Re: [Quickfixj-users] Do you validate?

[Philip W. <ph...@wh...>]

We turn off some validation. For instance AllowUnorderedGroupFields tends to be useful for some FIX Engines that just don’t order group fields :/

Personally what I’d like to start doing more is replaying data through a test system with more stringent validation to allow us to accept tags we don’t know but also to become more aware of it and so we can update our dictionaries (and work towards standardisation with our partners)

If we reject a message obviously that’s a monitoring event that we pick up and investigate.

I think accepting any message is dangerous and sort of leads to not really having a spec at all. Postel’s principle has serious downsides when trying to work out what you actually support. It also might mean, for example, filling an order incorrectly.

It is far better if you can come to an agreement that FIX spec changes are something that should be tested and notified in advance. Or at least a principle of how to handle tags not defined in the spec so as to handle the bad fill case.

Best,

Philip Whitehouse

> On 1 Jul 2018, at 20:37, <eri...@th...> <eri...@th...> wrote:
> 
> QuickFIX/J Documentation: http://www.quickfixj.org/documentation/
> QuickFIX/J Support: http://www.quickfixj.org/support/
> 
> 
> If repeating groups are being changed you have more than a validation problem, you have a potential parsing problem. The parser needs to know what belongs in a group or it can get confused about when the last group in a set ends. It’s one of the lovely side effects of FIX’s refusal to add another field separator for groups. If the counterparties are adding tags to repeating groups, you need the new dictionary.
> 
> That said, the only reason to not turn off validations is if you really, honestly, for sure, want to reject a message that can be parsed but doesn’t meet the spec. Personally, I’ve yet to deal with someone that really wanted that. 
> 
> I wish they defaulted to off, TBH.
> 
> 
>> On Jul 1, 2018, at 11:55, Robert Nicholson <rob...@gm...> wrote:
>> 
>> 
>> 
>> So my FIX clients typically process execution reports from ISVs and it seems a lot of them lately are beginning to change their FIX dictionary more often and if we do not upgrade
>> with them we typically end up with a new repeating group tag added that will fail validation.
>> 
>> So, is there any setting to turn off just the tag occurs more than once error or does that require disabling all validation?
>> 
>> I’m really tempted to turn off validation since the addition of a group tag that I don’t care about is something I can live with rather than failing the whole validation.
>> 
>> 
> 
> ------------------------------------------------------------------------------
> Check out the vibrant tech community on one of the world's most
> engaging tech sites, Slashdot.org! http://sdm.link/slashdot
> _______________________________________________
> Quickfixj-users mailing list
> Qui...@li...
> https://lists.sourceforge.net/lists/listinfo/quickfixj-users