Once in a while there are these intriguing situations that trigger me, as they play into what I always call "professional pride", or lack thereof. For example, in order to administer your Atlassian Forge app you can go to a developer console. In it, you can see a list of installations. But for some reason, if your app is popular enough, the filter options for that list are limited. As per the description of FRGE-1952 (https://lnkd.in/ebVYmivU) "To improve installations page load performance for apps with over 1,000 installs, we removed filters like license, edition, version, and sort functionality and provided download option. Users must enter the full site name, and page numbers are not clickable." Multi-facetted search is a "solved" problem. There are plenty of solutions for this (incl. open source), like ElasticSearch, OpenSearch and Solr. Atlassian has implemented these types of search engines in her core products. Deploying these solutions are often matter of days, not months, So how can we live in 2026 and have a Jira ticket to support multi-facetted searching for a broken product already shipped? How can a product manager believe that it is acceptable to ship a solution that limits the search UX instead of spending a few more sprints to deliver a scalable solution? What role does professional pride play in this? Do we truly believe that it is always better to ship as soon as possible? What good is it to the end-user to ship a feature that is unusable? At what point is it better to not have a feature than a shitty feature with an open-ended Jira ticket gathering dust in a backlog and a community post (https://lnkd.in/epxBuZWg) signalling yet another complaint of customer dissatisfaction? If we're not building for our customer, then who are we building for?
Paweł Indyk
2d
4 months ago I received same answer from Atlassian support. Could not believe their solution for low performance of the dev console... They just used the brute force and turned the license control into the nightmare. But hey, what we know right... Funny thing is that in the same message I got an info that some of the vendors still have access to those filters. What is the key here - no one said...
When you say this is a *solved* problem, do you mean there are patterns in production at Atlassian scale & architecture or larger that could definitively meet the need here? Do you have publicly available examples of said code? Or do you mean that theoretically this is a solved problems based on your knowledge of extrapolated problems and solutions? Not disagreeing, just truly curious!