Bugzilla

Actually not the regression I assumed above. Still may be a regression by something. It seems to be a threading issue maybe similar to bug 1801067. If I don't have this change in the mozilla code I see a one time assert crash and .eml attachments show as blank:
```
diff --git a/xpcom/base/nsISupportsImpl.cpp b/xpcom/base/nsISupportsImpl.cpp
--- a/xpcom/base/nsISupportsImpl.cpp
+++ b/xpcom/base/nsISupportsImpl.cpp
@@ -35,17 +35,19 @@ nsresult NS_FASTCALL NS_TableDrivenQI(vo
 
 #ifndef XPCOM_GLUE_AVOID_NSPR
 #  ifdef MOZ_THREAD_SAFETY_OWNERSHIP_CHECKS_SUPPORTED
 nsAutoOwningThread::nsAutoOwningThread() : mThread(PR_GetCurrentThread()) {}
 
 void nsAutoOwningThread::AssertCurrentThreadOwnsMe(const char* msg) const {
   if (MOZ_UNLIKELY(!IsCurrentThread())) {
     // `msg` is a string literal by construction.
+#if 0  // gds: stop crashing?
     MOZ_CRASH_UNSAFE(msg);
+#endif
   }
 }
 
 bool nsAutoOwningThread::IsCurrentThread() const {
   return mThread == PR_GetCurrentThread();
 }
 
 nsAutoOwningEventTarget::nsAutoOwningEventTarget()
```

Actually not the regression I assumed above. Still may be a regression by something. It seems to be a threading issue maybe similar to bug 
1801067. 
**Edit:** I was wrong, he code below has no effect on this bug.
If I don't have this change in the mozilla code I see a one time assert crash and .eml attachments show as blank:
```
diff --git a/xpcom/base/nsISupportsImpl.cpp b/xpcom/base/nsISupportsImpl.cpp
--- a/xpcom/base/nsISupportsImpl.cpp
+++ b/xpcom/base/nsISupportsImpl.cpp
@@ -35,17 +35,19 @@ nsresult NS_FASTCALL NS_TableDrivenQI(vo
 
 #ifndef XPCOM_GLUE_AVOID_NSPR
 #  ifdef MOZ_THREAD_SAFETY_OWNERSHIP_CHECKS_SUPPORTED
 nsAutoOwningThread::nsAutoOwningThread() : mThread(PR_GetCurrentThread()) {}
 
 void nsAutoOwningThread::AssertCurrentThreadOwnsMe(const char* msg) const {
   if (MOZ_UNLIKELY(!IsCurrentThread())) {
     // `msg` is a string literal by construction.
+#if 0  // gds: stop crashing?
     MOZ_CRASH_UNSAFE(msg);
+#endif
   }
 }
 
 bool nsAutoOwningThread::IsCurrentThread() const {
   return mThread == PR_GetCurrentThread();
 }
 
 nsAutoOwningEventTarget::nsAutoOwningEventTarget()
```

Actually not the regression I assumed above. Still may be a regression by something. It seems to be a threading issue maybe similar to bug 1801067. 
**Edit:** I was wrong, he code below has no effect on this bug.
If I don't have this change in the mozilla code I see a one time assert crash and .eml attachments show as blank:
```
diff --git a/xpcom/base/nsISupportsImpl.cpp b/xpcom/base/nsISupportsImpl.cpp
--- a/xpcom/base/nsISupportsImpl.cpp
+++ b/xpcom/base/nsISupportsImpl.cpp
@@ -35,17 +35,19 @@ nsresult NS_FASTCALL NS_TableDrivenQI(vo
 
 #ifndef XPCOM_GLUE_AVOID_NSPR
 #  ifdef MOZ_THREAD_SAFETY_OWNERSHIP_CHECKS_SUPPORTED
 nsAutoOwningThread::nsAutoOwningThread() : mThread(PR_GetCurrentThread()) {}
 
 void nsAutoOwningThread::AssertCurrentThreadOwnsMe(const char* msg) const {
   if (MOZ_UNLIKELY(!IsCurrentThread())) {
     // `msg` is a string literal by construction.
+#if 0  // gds: stop crashing?
     MOZ_CRASH_UNSAFE(msg);
+#endif
   }
 }
 
 bool nsAutoOwningThread::IsCurrentThread() const {
   return mThread == PR_GetCurrentThread();
 }
 
 nsAutoOwningEventTarget::nsAutoOwningEventTarget()
```