RR auto-restart after plug-in error

I've encountered something that I can't find an answer to.

Every now and then when a plug-in (All in C# .Net 3.5) fails during testing (especially with an uncaught error), Riderunner shuts-down and then restarts all by itself.

Is this the expected behavior?

Is there a RR.ini parameter or other way to prevent the restart?

Could it be something in the plug-in? To the best of my knowledge they all follow the format/rules given in the examples. The case in point is a plugin the only accepts a command and returns code (2) and there was an error in the plug-in internal parsing of the command. Other than that, at this point in development, it does nothing else.

This is a problem for me in that upon this automatic restart, all of my log files are overwitten and other nasty things happen, and worse yet, the restarted RR is not completely usable. Actually if the original error is persistent, I can only stop the restart cycle by killing the RR process tree using the task manager.

Anyone else encountered this?

Regards,
Jim

Well first of all, I would make sure the plugin doesn't have any uncaught errors.

RR doesn't restart itself for any reason, however, if any application/dll/script creates a reference to the SDK object it will cause RR to automatically load (if not already loaded). If your plugin is accessing/creating the SDK object to send commands/obtain info from RR on different parts of the application then it is possible for one part to cause the error/exception (causing RR to close) while another part could be referencing the SDK object causing RR to re-load. This may also happen on different levels like another unrelated application referencing the SDK object while your (or another) plugin crashed and caused RR to close. I haven't seen it happen because of two RR plugins but I would not rule it out, but it's definitely more likely with a separate EXE or Script (that won't crash with RR) and something internal to RR (i.e. bug or plugin) causing it to crash.

guino,

Thank you for a very helpful response (unlike some others). Since I have several .exe files running as well as plug-ins, I now have a path to correcting the problem. All of the exe files do "instantiate" the SDK and use it extensively to update/maintain RR variables and request other actions.

Clearly, if a plug-in has failed, causing RR to fail and an exe attempts to use the RR SDK interface during or after the RR failure, RR will be restarted. I think I will take several remedial approaches. One, check for RR running before all calls from the exe files to the SDK and two, a more "sledge-hammer" approach, to archive an existing log file as each component starts. At least then I can find what actually casued the failure in the first place.

Again thanks.

Regards,
Jim

So don't believe in catching the errors in your code in the first place?

Or being advised to do so helpful. Good to know.

Quote:

---

Originally Posted by Enforcer

So don't believe in catching the errors in your code in the first place?

Or being advised to do so helpful. Good to know.

---

Not even worth commenting on.

Regards,
Jim

This doesn't need to go further.

Regards,
Jim