aklefdal's tech blog

Password-less access from pipeline to database

Mon, 17 Mar 2025 00:00:00 +0000

The goal

The goal is to be able to run database migrations as part of a Azure DevOps Pipeline with using a stored secret anywhere. The databases are Azure Sql Databases, connected to a VNET, with no public access. There are two major problems to solve:

Network access
Authentication

Background

My current project is at a company with a software solution that has been in development for more than 20 years. Traditionally it ran on-premise, that is at the customer’s site. This is still the case for a few customers, but most customers are now using the service served from a cloud-hosted solution - Software-as-a-Service - SaaS

Generate a sequence of DateOnly in F#

Sun, 29 Dec 2024 00:00:00 +0000

For a hobby project I had to generate a sequence of DateTime (a.k.a. IEnumerable<DateTime>).

Generating a sequence from simple types is easy:

let seqOfInt (start: int) (end: int) =
    seq { start .. end }

But we cannot use range operator with DateOnly:

// Not valid code!
let badSeqOfDateTime (startDate: DateOnly) (endDate: DateOnly) =
    seq { startDate .. endDate }

In my first attempt at this (and current code as of today), I ended up with a while loop and a mutable:

Query Azure Table Storage using F#

Sun, 29 Dec 2024 00:00:00 +0000

When looking at the signature of the TableClient.QueryAsync method in Azure.Data.Tables SDK, it is not clear how to use it. It returns a AsyncPageable<T>. Fortunately that again implements the IAsyncEnumerable<T>, which means that we can use the Fsharp.Control.TaskSeq library to get the resulting entities.

Install FSharp.Control.TaskSeq NuGet-package to your project.
open FSharp.Control
Write your code using the TaskSeq module

let getPartitionRows (tableClient: TableClient) (partition: string) =
    task {
        return!
            tableClient.QueryAsync<TableEntity>(filter = $"PartitionKey eq '{partition}'")
            |> TaskSeq.map fromEntity
            |> TaskSeq.toArrayAsync
    }

This is written as a reminder to myself, so that I can easily find it again

Great improvement of build times

Sun, 30 Dec 2012 00:00:00 +0000

The two last days I presented the organization of our solution, and did some improvements.

Next up in the refactorings is to move the service implementations to the same project as the services themselves. They will always be deployed together, so there is no need to keep them in separate assemblies. Each service implementation project contains only one small class, so this is an easy refactoring.

This shaved off another 9 projects and 5-10 seconds of the build time, and we are down to 65-75 seconds.

Moving code to common projects

Sat, 29 Dec 2012 00:00:00 +0000

Yesterday I talked about our current setup, Today I will start to refactor away projects.

First up is to combine all the ServiceContracts-projects to one common ServiceContract-project

This reduced the build time from 90-100 seconds to 80-90 seconds. Not much, and maybe not enough to be statistically significant.

Next up was to combine all the Services-projects to one common Services project.

Painful development cycle – rebuild and run unit tests in 4 minutes

Fri, 28 Dec 2012 00:00:00 +0000

This Christmas I’m going try out improving the development environment at my current project. The baseline is as follows.

The project is big and has been running for years. Previously they have been working on the parts of the systems that the customers use through a web solutions, and the parts that the customer-facing employees use.

The part I am working on now was started fairly recently, and consists of the core system, which mainly run in the background using batch jobs, and providing services for the other parts of the system. This is what I will be talking about here.

What is ACM?

Fri, 22 Jun 2012 00:00:00 +0000

June 6 the company I work for, Computas, was selected as Gold Winner in Global Awards for Excellence in Adaptive Case Management hosted by the Workflow Management Coalition

ACM stands for Adaptive Case Management. It is an evolution of BPM (Business Process Management), or maybe a response to that concept. Its foundation is process management for knowledge workers. The characteristics of knowledge work is that it is unpredictable, and that you select what to do based on your knowledge, not on a predefined process.