We still had about a dozen developers combined on FogBugz+Kiln at the very end.
The plug-in API was very bad for performance and security, especially in a multitenant cloud application. That led a team to experimentally reimplement the entire frontend as a single-page app (code named Project Ocelot, whose shirt I am wearing today). Webhooks and a well-designed web API would have been a much tidier solution than the plug-in API.
FogBugz For Your Server had constant support costs, like Ben said, but also the application started growing all sorts of supporting services (a QueueService, an ElasticSearch cluster, Redis…) that made the existing InstallerShieid installer a huge cost to maintain.
Later, I wrote an “autosetup” script in PowerShell to help developers and support engineers onboard faster, which later became the new installer for FogBugz On-Site / Manuscript On Premises. It was designed for a single edition of Windows (Server 2012 iirc?) and SQL Server. We sold a million bucks worth of licenses for that.
The plug-in API was very bad for performance and security, especially in a multitenant cloud application. That led a team to experimentally reimplement the entire frontend as a single-page app (code named Project Ocelot, whose shirt I am wearing today). Webhooks and a well-designed web API would have been a much tidier solution than the plug-in API.
FogBugz For Your Server had constant support costs, like Ben said, but also the application started growing all sorts of supporting services (a QueueService, an ElasticSearch cluster, Redis…) that made the existing InstallerShieid installer a huge cost to maintain.
Later, I wrote an “autosetup” script in PowerShell to help developers and support engineers onboard faster, which later became the new installer for FogBugz On-Site / Manuscript On Premises. It was designed for a single edition of Windows (Server 2012 iirc?) and SQL Server. We sold a million bucks worth of licenses for that.