Pinned Tweet
Welcome to BugSplat 🐛 Learn how we can help you more quickly and easily find and fix critical defects causing issues in your app --> bugsplat.wistia.com/medias/3…
Most teams add crash reporting after launch, when angry Discord messages are already piling up and you're stuck reverse-engineering "it crashed" reports with no debugger in sight.
Add crash reporting to your game before the first external tester runs a build. 🧵
1
17
The trigger is your first external tester: a closed alpha, a friends-and-family build, a publisher milestone where someone outside the team runs the game.
You can't attach a debugger to a playtester's PC. All you get is "it crashed" and maybe a vague description.
1
1
The article walks through the three things to set up before testers show up: the SDK integrated into the build, symbol files uploaded from the build you hand out, and one forced test crash.
blog.bugsplat.com/when-to-ad…
2
Capturing a crash dump? A day or two of work. Capture is the easy 20% of crash reporting.
The build-or-buy decision is about the other 80%: grouping duplicate crashes, symbolicating stack traces, tracking crash rates across versions, alerting the right people. 🧵
1
1
7
An engineer with Cursor or Claude Code can scaffold a crash collection endpoint in an afternoon. The hard part is domain knowledge: grouping crashes with slightly different stack traces but one root cause, or symbolicating when a build pipeline emits symbols in three formats.
1
10
Our article lists the requests that follow an in-house prototype (crash rates by version, email alerts, a Jira integration, symbol servers) and makes the case for buying a crash reporting tool.
blog.bugsplat.com/build-vs-b…
4
0xC0000005, EXCEPTION_ACCESS_VIOLATION, is the most common way a C++ program dies. Before you open the debugger, read the faulting address. A small address and a large address come from two different bugs. 🧵
1
110
A small address such as 0x1C means a null pointer: your code reached for a field 28 bytes into an object that was null.
A large or random-looking address usually means a wild or dangling pointer, memory that was freed or went out of scope and is still being used.
1
8
On a user's machine you have no debugger. A crash report with line numbers and local variables shows that inventory was null on line 214. Our article lists the usual causes.
blog.bugsplat.com/access-vio…
4
A user hits a crash you already fixed two releases ago. They file a support ticket. You explain "update to the latest version."
A BugSplat support response shows the upgrade message in the user's browser right after the user submits the crash report. 🧵
1
1
23
The default support response is displayed when a crash report has no stack key-specific message. Configure the default with the Edit Default Response button on the Support Response page.
A support response can also be branded with your company or application logo.
1
1
4
Our docs cover creating a crash-specific support response, setting the default response and localizing support responses with the application key.
docs.bugsplat.com/guides/pro…
1
3
Nothing pulls a player out of your game faster than a crash. You can't undo the crash, but you do control what the player sees next.
BugSplat for Windows 9.0 includes a new crash dialog that you can style to match your game. 🧵
1
19
Upgrading a C++ app? Add BugSplatReporter.exe to your installer. If BugSplatReporter.exe is missing, crash reports still upload but your players never see the dialog.
On .NET, the BugSplat NuGet package copies BugSplatReporter.exe for you.
1
5





