[DMS-92] Basic eTag creation #381
Conversation
| @@ -20,6 +21,10 @@ public async Task Execute(PipelineContext context, Func<Task> next) | |||
| DateTimeOffset utcNow = DateTimeOffset.UtcNow; | |||
| string formattedUtcDateTime = utcNow.ToString("yyyy-MM-ddTHH:mm:ssZ"); | |||
| context.ParsedBody["_lastModifiedDate"] = formattedUtcDateTime; | |||
| context.ParsedBody["_etag"] = DateTime | |||
There was a problem hiding this comment.
I mimic'd the way ODS API generates etag.
There was a problem hiding this comment.
I do wonder if it would be better to generate etags from a hash of the document. Since DMS could be running on multiple servers, there's a teeny-tiny chance the clocks are out of sync in just the right way for different documents to have the same timestamp. Just something to think about.
| using EdFi.DataManagementService.Core.Pipeline; | ||
| using Microsoft.Extensions.Logging; | ||
|
|
||
| namespace EdFi.DataManagementService.Core.Middleware | ||
| { | ||
| internal class InjectLastModifiedDateToEdFiDocumentMiddleware(ILogger _logger) : IPipelineStep | ||
| internal class InjectVersionMetadataToEdFiDocumentMiddleware(ILogger _logger) : IPipelineStep |
| namespace EdFi.DataManagementService.Core.External.Model; | ||
|
|
||
| // A string type branded as an Etag which uniquely identifies a specific version of a document | ||
| public record struct Etag(string Value); |
| IUpsertRequest; | ||
| IUpsertRequest | ||
| { | ||
| } |
There was a problem hiding this comment.
Do we need this change?
| @@ -16,7 +16,8 @@ public record UpsertResult | |||
| /// A successful upsert request that took the form of an insert | |||
| /// </summary> | |||
| /// <param name="NewDocumentUuid">The DocumentUuid of the new document</param> | |||
| public record InsertSuccess(DocumentUuid NewDocumentUuid) : UpsertResult(); | |||
| /// <param name="Etag">The Etag of the new document</param> | |||
| public record InsertSuccess(DocumentUuid NewDocumentUuid, Etag Etag) : UpsertResult(); | |||
There was a problem hiding this comment.
Does a backend really need to return it? Doesn't core set it?
simpat-adam
force-pushed
the
DMS-92
branch
3 times, most recently
from
5fb2e75 to
aa177b2
Compare
Definitely could be done. Couple thoughts
|
Good points. Let's just keep it in the back of our minds. |
No description provided.