2 posts tagged with "postgresql"

If you’ve spent any time around code, you’ll know that // TODO comments are a fact of life. They're a common pattern, regardless of how you feel about them! For instance, the Kubernetes codebase has over 2,000, and Linux has over 3,000.

In the MergeStat codebase, we use them as a low-effort way to track small, technical debt items. Sometimes, a future fix or refactor might not warrant a ticket (it’s too small and too specific to a piece of implementation detail), but it’s still worth making a note. TODO comments are a good fit for this because they are:

• Low effort - very easy to add and remove, just leave a comment in the code
• Safe from context switching - no need to switch into a ticketing system, stay in the editor and near the relevant code
• Tracked in version control - so you have a history and audit trail

We recently connected with Ivan Smirnov, CTO of Elude on this topic, and were excited to learn about his enthusiasm for tracking TODO comments 🙂. He shared that during his time at Google, there was an internal system which aggregated TODOs across codebases, as a way of surfacing parts of code worth returning to. He missed having a similar solution in his role at Elude.

Luckily, we were able to help with a MergeStat + Grafana based solution! Elude operates a self-hosted instance of MergeStat (using our Kubernetes helm chart), and connects to its PostgreSQL database with a Grafana instance. We collaborated on putting together a starting “TODO Tracker” dashboard, which is available in our examples as a Grafana export:

"Elude currently uses TODO comments as a low friction mechanism to track technical debt. MergeStat is the missing link that allows us to turn these comments into trackable, actionable tasks." – Ivan Smirnov, Elude CTO

The SQL involved looks something like this:

SELECT
    git_blame.line,
    git_blame.author_name,
    git_blame.author_email,
    git_blame.author_when,
    REPLACE(repos.repo, 'https://github.com/', '') AS repo,
    repos.repo || '/blob/main/' || git_blame.path || '#L' || git_blame.line_no AS url -- generate a link to the line in GitHub
FROM git_blame
INNER JOIN repos ON repos.id = git_blame.repo_id
WHERE git_blame.line LIKE '%TODO%'
ORDER BY git_blame.author_when ASC

and should be fairly easy to customize to different needs:

• Only apply to certain repos
• Filter out certain file paths by pattern
• Look for FIXME and BUG comments as well
• Parse out "assignees" (i.e. TODO(patrickdevivo))
• etc...

If you're interested in taking a look at your own // TODO comments, go ahead and try out a local instance of MergeStat!

Today we're excited to announce the availability of in-app SQL querying in the latest versions of MergeStat! 🎉

One of the more significant pain points noted by our early adopters has been the need for external tools to access the data MergeStat provides, even for simple, one-off queries. Many users (ourselves included) lean on existing data products such as Grafana, Metabase and Superset to consume MergeStat data (in dashboards, alerts and reports).

With the addition of our query interface (in the Queries tab of our management console), MergeStat users can now execute SQL directly in our app.

Query results can be copied or downloaded as CSV or JSON, for quick use in other tools and contexts. We believe this is a significant step-forward in our app's functionality, and will continue to invest in this area of our management console. Keep a lookout for additional features, including:

• Saved and examples queries
• Inline schema documentation and editor auto-completions
• Basic charting and data visualization
• Query execution history

It's important to note that as always, the MergeStat PostgreSQL database can be connected directly to by BI tools, desktop SQL clients, or SQL drivers.

Check out the latest MergeStat release to get started!