Crash reports
Erik says he doesn’t send crash reports to Apple. That’s unfortunate. Unless there’s highly confidential information in your crash log — say, you’re working on a new application that you don’t want anyone at Apple to know anything about — what’s the harm? It’s much easier for Apple to determine whether a particular crash log shows up multiple times across millions of users (by aggregating the data) than for any one user to make that judgment on his own. And sure, Apple may conclude that it’s a bug in the application and not in the system frameworks, but it’s a shame if anyone wants to prevent Apple from making that determination in the first place.
In other words, if you want to make Mac OS X a better product, send in those crash logs. Hitting cancel instead of sending them in doesn’t help anyone.
Can Sar Said,
March 13, 2005 @ 7:49 pm
I for one didn’t know that Apple actually looked at that stuff (it’s aggregation, obviously they wouldn’t look at individual one’s) and friends working at Apple had told me that they doubt that sending it would help in any way. That dialogue box always gave me a creepy Microsoftish-feeling, but now I’m reconsidering that view.
Darth Sidious Said,
March 18, 2005 @ 2:28 pm
Hum, IMHO, it’s pretty useless submitting JUST the info of the crash.
If you don’t provide information on the context, on the way it can be reproduced, then it’s most of the times difficult to reproduce the problem.
Darth Sidious Said,
March 18, 2005 @ 2:32 pm
Hum, and just thinking about it, software context might not be enough in some cases.
This week, someone reported me a bug which could be described as :” The Control + Shift + T short-cut does not work with the OS version I’m using”.
It turned out after spending time trying to reproduce the bug on my machine that the Control key was broken on his keyboard…
Eric Said,
March 18, 2005 @ 2:33 pm
When I get a crash log, I tend to ignore the context information unless it makes it very clear that the bug is completely reproducible. In a lot of cases you can actually figure out what happened just by analyzing the crash log and looking at the source code, even if you can’t reproduce it.