The problem of repeated crashes

Postmortem app crash reporting and analysis is a bit of a hobbyhorse for me. I see it as an extension of the bug reporting facility - a crash in production (typically, but not always) indicates a bug. But does every crash indicate a separate, distinct bug?

Project vs. collection

Some time ago, I've discussed the identity of the account that TFS agent jobs are using to connect back to TFS via the distributedTask PowerShell context variable, or the equivalent System.AccessToken release variable. Back at the time, I've concluded that the identity behind that token was "Build Service (CollectionName)". Well, just today I've encountered a case where a release was running under a different identity - "ProjectName Build Service (CollectionName)". Both coexist.

Abusing COM, this time in C#

Some time ago, I've written up a technique for setting up object oriented, call based interprocess communication using an idiosyncratic subset of COM. A case for something similar came up for me again, but the project is in C# with .NET 6. So I've adapted the same technique to work in a managed world.

Sequential vs. semantic

This is about the naming of releases in Azure DevOps (formerly known as TFS and VSTS). The default pattern of release names has been, as long as I can remember, sequential within the definition's scope: "Release 1", "Release 2", etc. That's been a minor pet peeve of mine for a while, and now we've come up with a technique to change this format to something more informative. Specifically, one can name releases with respect to the current product version, as every reasonable project team would want to.

OAuth scopes in Azure DevOps 2020

Here's another OAuth scope dump from the latest on-prem instance of Azure DevOps.

The same caveat applies: some of those might be cloud only.

It's AzDevOps all the way down

Oy vey. Some posts ago, I've mentioned that all TFS Azure DevOps collection databases had the same schema, haven't I? Well, not quite.

Full circle

Following up from my last post about generating a CSR, I went through the motions of testing the full cycle for said CSR. I'd sign it with a bogus (but locally trusted) authority certificate, install the signed cert locally, export the PFX, install it on a Web server, select the cert for TLS on an IIS website, and hit it with a bunch of desktop browsers. The certificate works as expected.