Passing strings that look like dates from orquestator to activity changes the time zone

[afrancoc2000]

Describe the bug
I have an orchestator that sends an object to an activity that includes strings that look like dates, from localhost (-05:00) if I send a date like '2021-03-15T16:00:00-07:00' it gets changed to '2021-03-15T18:00:00-05:00'.

Investigative information

• Durable Functions extension version: 2.4.1
• durable-functions npm module version: ^1.4.6
• Language (JavaScript/TypeScript) and version: typescript ^3.3.3
• Node.js version: 10.23

If deployed to Azure App Service

• Timeframe issue observed: 2021-03-12T10:00:00-05:00 - 2021-03-12T23:00:00-05:00
• Function ID: Id=2b8e96e0-2ba6-499c-9d87-f6ec1f41db2b
• Region: East US
• Orchestration instance ID(s): I don't know

To Reproduce
Steps to reproduce the behavior:

Create an orchestator and an activity

Orchestator

const orchestrator = df.orchestrator(function* (context) {
        const message = [[{
        "fechaMedida":"2021-03-15T16:00:00-07:00",
        "tipoMedicion":"medida"
    }]]
    yield context.df.callActivity('generar-publicar', message);
});

export default orchestrator;

Activity:

const activityFunction: AzureFunction = async function (context: Context): Promise<void> {
    const messages = context.bindings.messages;
    context.log(`messages: ${JSON.stringify(messages)}`);
};

export default activityFunction;

Check the output in the log, the string changes to the time zone of your machine on localhost and to UTC in the server.

messages: [[{"fechaMedida":"2021-03-15T18:00:00-05:00","tipoMedicion":"medida"}]]

Expected behavior
An output log like this:

messages: [[{"fechaMedida":"2021-03-15T16:00:00-07:00","tipoMedicion":"medida"}]]

@davidmrdavid Here is the issue, hope it helps.

For now I am using JSON.stringfy() on the orchestator and JSON.parse on the activity to mitigate the problem.

Thanks.

[cgillum]

@davidmrdavid are you familiar with this issue? I feel like I've seen it mentioned before.

[davidmrdavid]

@cgillum: Yes, this is a known issue that requires changes in the node-worker to completely fix. So far, I have some internal mitigations that are able to get around this, but they're not perfect. I'm looking to merge a node-worker patch soon. However, this specific error instance should be handled by pre-existing mitigations.

Hi @afrancoc2000! Thank you so much for filing this issue. I'll add it to our current sprint and look into it as soon as tomorrow morning PDT. I'll get back to you!

[afrancoc2000]

@cgillum: Yes, this is a known issue that requires changes in the node-worker to completely fix. So far, I have some internal mitigations that are able to get around this, but they're not perfect. I'm looking to merge a node-worker patch soon. However, this specific error instance should be handled by pre-existing mitigations.

Hi @afrancoc2000! Thank you so much for filing this issue. I'll add it to our current sprint and look into it as soon as tomorrow morning PDT. I'll get back to you!

Thanks!