Posts

Golang Append

While working on some code base, I found out a weird quirk of append built-in function in Golang.

There was a string slice defined and after some operations, the function was returning two things, a string and a string slice.
... s := []string{"a", "b", "c", "d"} ... return s[1], append(s[:1], s[1+1:]...) The output is supposed to be "b" and "a", "c", "d" but what I got was "c" and "a", "c", "d". At first I thought I have discovered a bug but after reading the documentation it became clear that it is intentional. Turns out it has to do with implementation of Slice described in this blog. So, append modifies the underlying array and since arguments are evaluated from right to left, append is executed first which modified the underlying array to become "a", "c", "d" and then s[1] returns the 1st element of the modified array "c"…
We had a fun day at work today. Some new people joined and we planned a challenging treasure hunt in WeWork building for them. Then families of workmates joined us for a logo designing competition. We then played dumb charades and ate Pizzas together.
I finally bought the Interpreter Book on Kindle store.
A nice write-up about following RFC like process for managing Engineering team at organizations.

https://buriti.ca/6-lessons-i-learned-while-implementing-technical-rfcs-as-a-management-tool-34687dbf46cb

Vanity Imports in Golang

The imports in Golang are the fullpath of the package including the domain name. It is common to host the packages on Github and use it for import as well. However, there are two problems with it.
Long imports: Packages under Github have to have github.com/<username>/<repository-name> at the beginning which is rather long considering some repositories have nested sub-directories. This is true for internal packages in the repository as well, when you want to import a package from a different sub-directory but you still have to use the entire Github prefix. Hassle in switching the hosting The package name is the unique identifier for the package and having the hosting provider name in it means its a hassle to switch provider even for internal packages.
The two aforementioned problems can be solved using vanity imports in Go. What vanity import means is using a custom domain for package imports.
github.com/<username>/<repository>/pkg/<xyz> -> <user.t…

Backup SQLite

The Writefreely instance running the current website is using SQLite database to store the content. SQLite is an embedded relational database which primarily ships as a library with bindings for a lot of programming languages. It also ships a helper command line utility sqlite3 which can be used to interact with database file. This tool can be used to take backup by calling .backup command.
$ cat backup.sh !#/bin/sh sqlite3 /path/to/writefreely.db ".backup /backup/path/backup.db" $ This is equivalent to SQL dump which can be imported in a new instance if need arises. The backup process can be configured to run everyday using cron daemon. The cron expression for taking backup at midnight 12 is 0 0 * * *. The following line can be added to cronjob to automate the process. 0 0 * * * ~/backup.sh Cron expressions for different time/frequency can be easily generated using the web-based tool called crontab.guru.
For offsite backups, the SQL dump can be copied to a remote machine usi…

Golinks

This project is a very simple webapp inspired by Google's internal “go” system. First time I came to know about this was from this Gitlab thread. Also from the same thread I came to know about an open-source implementation of it at kellegous/go.

I have been working with Golang extensively at work and otherwise for the past year but this is my first web application in Go. The application is analogous to a URL shortner with the different that you get to choose the slug. It implements the functionality of adding and deleting as well as listing all links. I also semi-implemented authentication where it assumes an external system (middleware) to do it and set a specific HTTP header. I am using Cloudflare Access for implementing this in my current setup.

I used Mux router to build the endpoints with two middlewares for logging and authentication. For logging, I used a leveled logging library by Netflix engineer called zerolog. I used Go implementation of LevelDB which is an embedded …