I’ve opened a pull request for this feature here: 
https://github.com/apache/avro/pull/1276 
<https://github.com/apache/avro/pull/1276>

It’s succinct. If someone would kindly have a review and let me know their 
thoughts, it would be greatly appreciated!

Best,
Andreas | www.andreashailu.com <http://www.andreashailu.com/>




> On Jun 20, 2021, at 10:12 AM, Andreas Hailu <[email protected]> wrote:
> 
> I was able to get something working locally that supports this kind of 
> feature while staying true to the fluent-style of the SchemaBuilder. A 
> snippet from a test I wrote to see what it looks like:
> 
> SchemaBuilder.record("ValidationRecord")
>   .fields()
>   .name("IntegerField")
>   .notValidatingDefaults()
>   .type("int")
>   .withDefault(“foo”)
>   .endRecord();
> 
> Thoughts? If this is something we’d like to support, I can open a pull 
> request for review.
> 
> Best,
> Andreas | www.andreashailu.com <http://www.andreashailu.com/>
> 
> 
> 
> 
>> On Jun 1, 2021, at 11:12 AM, Andreas Hailu <[email protected] 
>> <mailto:[email protected]>> wrote:
>> 
>> Fixed a typo inline in the previous message - apologies.
>> 
>> 
>>> On Jun 1, 2021, at 11:07 AM, Andreas Hailu <[email protected] 
>>> <mailto:[email protected]>> wrote:
>>> 
>>> Hello,
>>> The SchemaBuilder's createFields methods pass a value of 'true' to the 
>>> validateDefault option within the Field constructor. While this is very 
>>> useful to have, it results in not being able to seamlessly migrate from 1.8 
>>> to >= 1.9 if an application/schema registry already has schemata with 
>>> invalid defaults.
>>> 
>>> Given that Schema.Parser allows us to configure this, what are your 
>>> thoughts on supporting the same idea within SchemaBuilder? I created a JIRA 
>>> [1] with some details and would be happy to contribute if we’d like to 
>>> support this.
>>> 
>>> [1] https://issues.apache.org/jira/browse/AVRO-3144 
>>> <https://issues.apache.org/jira/browse/AVRO-3144> 
>>> 
>>> Best,
>>> Andreas | www.andreashailu.com <http://www.andreashailu.com/>
>>> 
>>> 
>>> 
>>> 
>> 
> 

Reply via email to