Apologies if this would be a better fit for the user mailing list, but I
was unsure where to ask questions on (yet) unreleased functionality. I was really happy when I realized that FLINK-20334 <https://issues.apache.org/jira/browse/FLINK-20334> was adding Path Templating, but it seems to me that the possibility was not added to the Data Stream API. I may be misunderstanding the codebase, but in SerializableHttpFunctionProvider#48 the HttpFunctionProvider is always instantiated with an empty map of perNamespaceEndpointSpecs, which would be easy to change in the StatefulFunctionDataStreamBuilder. Is there a reason this wasn't considered? Is there anything tricky that is blocking this addition, or maybe path templating is not relevant for the DataStream API and I'm not seeing the most appropriate way to achieve the same goal? Thanks, Miguel |
This is the right place to ask such questions; in fact I'd suggest to
just open a JIRA ticket. On 2/18/2021 2:50 PM, Miguel Araújo wrote: > Apologies if this would be a better fit for the user mailing list, but I > was unsure where to ask questions on (yet) unreleased functionality. > > I was really happy when I realized that FLINK-20334 > <https://issues.apache.org/jira/browse/FLINK-20334> was adding Path > Templating, but it seems to me that the possibility was not added to the > Data Stream API. I may be misunderstanding the codebase, but in > SerializableHttpFunctionProvider#48 the HttpFunctionProvider is always > instantiated with an empty map of perNamespaceEndpointSpecs, which would be > easy to change in the StatefulFunctionDataStreamBuilder. > > Is there a reason this wasn't considered? Is there anything tricky that is > blocking this addition, or maybe path templating is not relevant for the > DataStream API and I'm not seeing the most appropriate way to achieve the > same goal? > > Thanks, > Miguel > |
Hi Miguel,
As Chesnay mentioned already, asking about the direction of some ongoing developed feature in the dev@ list is absolutely fine and correct. And it's indeed a good idea to open a JIRA ticket for this. As a quick pointer, I don't think there is any specific reason why path templating wasn't exposed in the DataStream API - this should be possible with some rework of the RequestReplyFunctionBuilder. I think it was initially left out as it's quite an isolated extended change to the core development of the path templating feature. If you'd like to open a JIRA on this, we'd be happy to move the discussion of adding feature to the ticket. Cheers, Joanne On Fri, Feb 19, 2021 at 3:30 AM Chesnay Schepler <[hidden email]> wrote: > This is the right place to ask such questions; in fact I'd suggest to > just open a JIRA ticket. > > On 2/18/2021 2:50 PM, Miguel Araújo wrote: > > Apologies if this would be a better fit for the user mailing list, but I > > was unsure where to ask questions on (yet) unreleased functionality. > > > > I was really happy when I realized that FLINK-20334 > > <https://issues.apache.org/jira/browse/FLINK-20334> was adding Path > > Templating, but it seems to me that the possibility was not added to the > > Data Stream API. I may be misunderstanding the codebase, but in > > SerializableHttpFunctionProvider#48 the HttpFunctionProvider is always > > instantiated with an empty map of perNamespaceEndpointSpecs, which would > be > > easy to change in the StatefulFunctionDataStreamBuilder. > > > > Is there a reason this wasn't considered? Is there anything tricky that > is > > blocking this addition, or maybe path templating is not relevant for the > > DataStream API and I'm not seeing the most appropriate way to achieve the > > same goal? > > > > Thanks, > > Miguel > > > > |
Thanks, created FLINK-21420
<https://issues.apache.org/jira/browse/FLINK-21420>. Tzu-Li (Gordon) Tai <[hidden email]> escreveu no dia sexta, 19/02/2021 à(s) 08:43: > Hi Miguel, > > As Chesnay mentioned already, asking about the direction of some ongoing > developed feature in the dev@ list is absolutely fine and correct. > And it's indeed a good idea to open a JIRA ticket for this. > > As a quick pointer, I don't think there is any specific reason why path > templating wasn't exposed in the DataStream API - this should be possible > with some rework of the RequestReplyFunctionBuilder. > I think it was initially left out as it's quite an isolated extended > change to the core development of the path templating feature. > > If you'd like to open a JIRA on this, we'd be happy to move the discussion > of adding feature to the ticket. > > Cheers, > Joanne > > On Fri, Feb 19, 2021 at 3:30 AM Chesnay Schepler <[hidden email]> > wrote: > >> This is the right place to ask such questions; in fact I'd suggest to >> just open a JIRA ticket. >> >> On 2/18/2021 2:50 PM, Miguel Araújo wrote: >> > Apologies if this would be a better fit for the user mailing list, but I >> > was unsure where to ask questions on (yet) unreleased functionality. >> > >> > I was really happy when I realized that FLINK-20334 >> > <https://issues.apache.org/jira/browse/FLINK-20334> was adding Path >> > Templating, but it seems to me that the possibility was not added to the >> > Data Stream API. I may be misunderstanding the codebase, but in >> > SerializableHttpFunctionProvider#48 the HttpFunctionProvider is always >> > instantiated with an empty map of perNamespaceEndpointSpecs, which >> would be >> > easy to change in the StatefulFunctionDataStreamBuilder. >> > >> > Is there a reason this wasn't considered? Is there anything tricky that >> is >> > blocking this addition, or maybe path templating is not relevant for the >> > DataStream API and I'm not seeing the most appropriate way to achieve >> the >> > same goal? >> > >> > Thanks, >> > Miguel >> > >> >> |
Free forum by Nabble | Edit this page |