Conversation
|
Sorry with fever in bed and can't think straight |
|
@danielmarbach no worries, get well soon! Some more context in this comment of mine: |
|
Out of curiosity, wouldn't it be better to actually encode the header as the number type directly, rather than going through a string? I believe headers do support numbers as value directly (at least I recall banging my head against that for a Rebus.rabbitmq bug a while ago) Though I guess allocation-wise it won't matter much as it's either a string allocation or a boxing allocation? I'm asking mainly to learn, as there might be some delicious knowledge to be had :) Either way thank you for the extremely quick response time on the issue! |
Yep, it would be, but RabbitMQ and QPid stray from the AMQP 0.9.1 spec when it comes to unsigned long long values: https://www.rabbitmq.com/amqp-0-9-1-errata Notice that the spec specifies that ...but RabbitMQ does not work that way: |
882980d to
4506de7
Compare
|
@zlepper your suggestion made me consider an alternative.
Considering that it would take publishing one message per millisecond over the next 292,471,208 years to exceed |
long
Fixes issue raised in #1805 Encode publish sequence number in x-dotnet-pub-seq-no as a `long` instead of ASCII string, unless value is greater than `long.MaxValue`.
4506de7 to
b057ca2
Compare
There's always |
|
Just tested this master branch locally and it seems to be a much better solution. In our case a SAP pi po rest adapter is having a lot of trouble with the previous value of the x-dotnet-pub-seq-no header while transforming it to XML. @lukebakken any idea when a new (alpha) release will be made? Many thanks for this fix, a saved us from a big headache. |
As soon as I can fix this issue - #1802 It is difficult to reason about. |


Fixes issue raised in #1805