Stream: fhircast-github
Topic: fhircast-docs / Issue #353 Bas stu2 prepublish review
Github Notifications (FHIRcast) (Jan 04 2021 at 17:57):
isaacvetter commented on Issue #353:
Happy New Year, @bvdh !
Okay, of these four issues:
Presence of '-' in proprietary events - I suggest to file an issue on this and discuss it as part of STU3
Yes, will you please log this as a jira issue? (If one doesn't already exist?)
Subscription request format - last box related to websockets - "Required when hub.channel.type=websocket for re-subscribes and unsubscribes." is a duplication of the column that states it is conditional for webscockets. The sentence on its own is not wrong - remove it or make it an issue for STU3.
Will you please log this as a jira issue? I'm likely misunderstanding, because I'm uncertain you're right. Note that not only is
hub.channel.typerequired for websockets and not for webhooks, but more specifically is not required for initial subscribes when using websockets. Elsewhere, when we use CONDITIONAL, it typically means that websockets always requires and webhooks doesn't or vice-versa.
The computable syntax of FHIRcast events in section (https://github.com/HL7/fhircast-docs/blob/8200aa1a22201ab553f81a15bce6ddce02e3de86/docs/specification/STU2.md#fhircast-authorization--smart-scopes) does not correspond to the event definition. I suggest that we remove it, it has no impact on the spec and will prevent confusion.
In this section, there are two diagrams. The first is introduced with this phrase: "the FHIRcast syntax for workflow related events is:", and the second with: "The FHIRcast event name is also a computable syntax, the complete syntax for FHIRcast scopes is:". Yet, the first diagram describes only OAuth scope syntax and the second diagram describes the event name format within an OAuth2 scope. The text introducing the first image is wrong. Changed to: "the FHIRcast syntax for OAuth 2.0 access scopes is:".
Secondly, since we've also added an additional diagram describing EBNF for actual events (in the Event Definition Format hook name section). I'm also removing the full EBNF which attempts to document both the event name and OAuth2 scope EBNF structure.
There are still an issue I really do not agree with. It is related to events being stateless (https://github.com/HL7/fhircast-docs/blob/8200aa1a22201ab553f81a15bce6ddce02e3de86/docs/specification/STU2.md#events). This is not true. I cannot derive the current state of a session from a close event. I would like it to be so but the current spec works differently. The only way to make this happen is when only close events are allowed for currently open anchor resources. But this would go further than an editorial change. Therefore I suggest that we remove the sentence from the spec and create an issue on this to be resolved in STU3.
Bas, I think we're debating the meaning of words and not the intent or understanding. I updated the sentence: "FHIRcast events are stateless" to "FHIRcast events do not communicate previous state". Do you agree with this change? If not, let's remove the sentence.
Isaac
Github Notifications (FHIRcast) (Jan 04 2021 at 17:58):
isaacvetter edited a comment on Issue #353:
Happy New Year, @bvdh !
Okay, of these four issues:
Presence of '-' in proprietary events - I suggest to file an issue on this and discuss it as part of STU3
Yes, will you please log this as a jira issue? (If one doesn't already exist?)
Subscription request format - last box related to websockets - "Required when hub.channel.type=websocket for re-subscribes and unsubscribes." is a duplication of the column that states it is conditional for webscockets. The sentence on its own is not wrong - remove it or make it an issue for STU3.
Will you please log this as a jira issue? I'm likely misunderstanding, because I'm uncertain you're right. Note that not only is
hub.channel.typerequired for websockets and not for webhooks, but more specifically is not required for initial subscribes when using websockets. Elsewhere, when we use CONDITIONAL, it typically means that websockets always requires and webhooks doesn't or vice-versa.
The computable syntax of FHIRcast events in section (https://github.com/HL7/fhircast-docs/blob/8200aa1a22201ab553f81a15bce6ddce02e3de86/docs/specification/STU2.md#fhircast-authorization--smart-scopes) does not correspond to the event definition. I suggest that we remove it, it has no impact on the spec and will prevent confusion.
In this section, there are two diagrams. The first is introduced with this phrase: "the FHIRcast syntax for workflow related events is:", and the second with: "The FHIRcast event name is also a computable syntax, the complete syntax for FHIRcast scopes is:". Yet, the first diagram describes only OAuth scope syntax and the second diagram describes the event name format within an OAuth2 scope. The text introducing the first image is wrong. Changed to: "the FHIRcast syntax for OAuth 2.0 access scopes is:".
Secondly, since we've also added an additional diagram describing EBNF for actual events (in the Event Definition Format hook name section). I'm also removing the full EBNF which attempts to document both the event name and OAuth2 scope EBNF structure.
There are still an issue I really do not agree with. It is related to events being stateless (https://github.com/HL7/fhircast-docs/blob/8200aa1a22201ab553f81a15bce6ddce02e3de86/docs/specification/STU2.md#events). This is not true. I cannot derive the current state of a session from a close event. I would like it to be so but the current spec works differently. The only way to make this happen is when only close events are allowed for currently open anchor resources. But this would go further than an editorial change. Therefore I suggest that we remove the sentence from the spec and create an issue on this to be resolved in STU3.
Bas, I think we're debating the meaning of words and not the intent or understanding. I updated the sentence: "FHIRcast events are stateless" to "FHIRcast events do not communicate previous state". Do you agree with this change? If not, let's remove the sentence.
Isaac
Github Notifications (FHIRcast) (Jan 04 2021 at 18:11):
isaacvetter edited a comment on Issue #353:
Happy New Year, @bvdh !
Okay, of these four issues:
Presence of '-' in proprietary events - I suggest to file an issue on this and discuss it as part of STU3
Yes, will you please log this as a jira issue? (If one doesn't already exist?)
Subscription request format - last box related to websockets - "Required when hub.channel.type=websocket for re-subscribes and unsubscribes." is a duplication of the column that states it is conditional for webscockets. The sentence on its own is not wrong - remove it or make it an issue for STU3.
Will you please log this as a jira issue? I'm likely misunderstanding, because I'm uncertain you're right. Note that not only is
hub.channel.typerequired for websockets and not for webhooks, but more specifically is not required for initial subscribes when using websockets. Elsewhere, when we use CONDITIONAL, it typically means that websockets always requires and webhooks doesn't or vice-versa.
The computable syntax of FHIRcast events in section (https://github.com/HL7/fhircast-docs/blob/8200aa1a22201ab553f81a15bce6ddce02e3de86/docs/specification/STU2.md#fhircast-authorization--smart-scopes) does not correspond to the event definition. I suggest that we remove it, it has no impact on the spec and will prevent confusion.
In this section, there are two diagrams. The first is introduced with this phrase: "the FHIRcast syntax for workflow related events is:", and the second with: "The FHIRcast event name is also a computable syntax, the complete syntax for FHIRcast scopes is:". Yet, the first diagram describes only OAuth scope syntax and the second diagram describes the event name format within an OAuth2 scope. The text introducing the first image is wrong. Changed to: "the FHIRcast syntax for OAuth 2.0 access scopes is:".
Secondly, since we've also added an additional diagram describing EBNF for actual events (in the Event Definition Format hook name section). I'm also removing the full EBNF which attempts to document both the event name and OAuth2 scope EBNF structure.
There are still an issue I really do not agree with. It is related to events being stateless (https://github.com/HL7/fhircast-docs/blob/8200aa1a22201ab553f81a15bce6ddce02e3de86/docs/specification/STU2.md#events). This is not true. I cannot derive the current state of a session from a close event. I would like it to be so but the current spec works differently. The only way to make this happen is when only close events are allowed for currently open anchor resources. But this would go further than an editorial change. Therefore I suggest that we remove the sentence from the spec and create an issue on this to be resolved in STU3.
Bas, I think we're debating the meaning of words and not the intent or understanding. I updated the sentence: "FHIRcast events are stateless" to "FHIRcast events do not communicate previous state". _Do you agree with this change? If not, let's remove the sentence._
Isaac
Github Notifications (FHIRcast) (Jan 05 2021 at 10:06):
bvdh commented on Issue #353:
Allow for '-' to be present in proprietary FHIRcast events
==> https://jira.hl7.org/browse/FHIR-30210?filter=-2Subscription request format
==> https://jira.hl7.org/browse/FHIR-30211I updated the sentence: "FHIRcast events are stateless" to "FHIRcast events do not communicate previous state". Do you agree with this change? If not, let's remove the sentence.
-- I Agreee
Last updated: Apr 12 2022 at 19:14 UTC
