Remote

Darrel Miller

@darrel_miller@mastodon.social

mastodon 4.8.0-nightly.2026-10-06

HTTP Advocate, API Architect on Microsoft Graph, Editor of OpenAPI specification, Co-chair of IETF HTTPAPI working group. Architect for Kiota https://aka.ms/kiota

0 Followers

0 Following

12 Posts

Joined April 25, 2022

GitHub:

darrelmiller

Blog:

https://apievolution.tavis.ca

Darrel Miller @darrel_miller@mastodon.social

· 5mo ago

Replying to

@josepvives @nuclearplayer Sure it is technically a simple problem. But there are many processes in place that prevent that from happening. Accessing customer content is very tightly controlled. Privacy is something I care about and it is one of the reasons I chose to work at Microsoft rather one of the other big tech companies that do not have the same guardrails in place.

Darrel Miller @darrel_miller@mastodon.social

· 11mo ago

The A2A Agent card TypeScript definition has protocolVersion field that is required but has a default value. That is enough to make my head explode. But I need to translate that into proto3. If you have ever explored the saga of field presence in protobuf, you can probably imagine how I feel right now.

Darrel Miller @darrel_miller@mastodon.social

· 5mo ago

Replying to

@nuclearplayer When you call an API or make a git request to some remote repo, there are going to be logs of that activity on that remote site. We acknowledge that site owners need some visibility into what is happening on their service. However, when it comes to code that is downloaded and executed on a local machine there seems to be an expectation that the code owners no longer have any rights to see how that code is executing. Help me understand why the rules are different.

Darrel Miller @darrel_miller@mastodon.social

· 5mo ago

Replying to

@nuclearplayer I would think the important thing is what data is being collected, not the the fact that any data is being collected. If that remote site is collecting end user identifiable information, that should be as big a problem as if a local tool is doing it. What is good about a "source-open" collecting the telemetry is that you can see and verify what is being collected. You can't with a remote service.

Darrel Miller @darrel_miller@mastodon.social

· 5mo ago

Replying to

@nuclearplayer And as a Microsoft employee, my experience has been that we are extremely careful about not logging any information that directly identifies users and any customer created content. It isn't lip service to privacy. I've seen projects delayed while we scrub logs because a developer accidentally logged the name of some artifact that they should not have.

Darrel Miller @darrel_miller@mastodon.social

· 5mo ago

Replying to

@_aD @w I don't think anyone feels entitled. I think the product owners want to provide the best experience for their users and knowing how the product is used helps. For tools that primarily are clients for a backend service, then the service will know whenever a service call is made. I'm trying to fully understand the objection to capturing some additional usage information that doesn't make a service call. Is it the "slippery slope " problem?

Darrel Miller @darrel_miller@mastodon.social

· 5mo ago

Replying to

@josepvives @nuclearplayer Yeah. I can see that some random anecdote from a Microsoft employee is not any kind of assurance. It is frustrating from my perspective because we have to jump through hoops to get any kind of useful data to be "data driven" in our product work but from the outside world we are perceived to be partying on everyone's private data. More transparency would be good for everyone.

Darrel Miller @darrel_miller@mastodon.social

· 5mo ago

Replying to

@nuclearplayer @dalias the "legitimate interests" GDPR clause for pseudonymous information does seem to make this a grey area, but IANAL and I am not trying to make a judgement on what GitHub did. I'm trying to learn about the objections. I understand the desire for consent but we can see from the "accept cookie" mess that users can just be coerced to consent via fatigue. I wish we had a standardized opt-out mechanism like DNT tried to do.

Darrel Miller @darrel_miller@mastodon.social

· 9mo ago

In some ways, typing really was the blocker. It turns out that there are dozens of little "building block" libraries that I have always wanted to write, but just never found the time. Not really complex stuff. Just take a written specification that has done all the hard thinking and turn it into a library that meets my requirements.
Being able to turn specification words into packaged code with tests and docs in a trivially short amount of time, is actually a very handy thing.

Darrel Miller @darrel_miller@mastodon.social

· 3mo ago

Apparently Air Canada are having a systems outage. Passengers are unable to board because they don't seem to be able to load all the bags. I'm in YUL but apparently this affects everywhere.

Darrel Miller @darrel_miller@mastodon.social

· 10mo ago

Replying to

@timbray@cosocial.ca @eythian@teh.entar.net Technically you can, but technically it has no semantic meaning so you can't do anything useful with it. Any HTTP message can have a body. However only some methods allow the body to have meaning for the request. GET and DELETE don't, POST and QUERY do. It's a layering thing in the design.