Timeline for Responsibility to reproduce bugs
Current License: CC BY-SA 3.0
12 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Jun 14, 2013 at 15:33 | comment | added | Doc Brown | @BryanOakley: yes, I agree, and if I were the library owner, I would probably try to help any user observing problems to gather enough information for me so that I can reproduce the bug. For example, by adding additional logging fucntionality. Nethertheless I am very resistant against searching for bugs in the code without having a reproducible scenario first. | |
| Jun 14, 2013 at 11:14 | comment | added | Bryan Oakley | "must" is a pretty strong word. If I were the one to have written the library, I would definitely appreciate a test that reproduces it but I certainly wouldn't say they must provide it. The goal is to write good software as a team, there are no "musts" if everyone is cooperating. | |
| Jun 11, 2013 at 6:07 | comment | added | avakar | @DocBrown, oh, I wasn't opposing, you actually have my +1 :) I was correcting the "100% must provide repro steps", it's like 80% for us. | |
| Jun 10, 2013 at 16:21 | comment | added | Doc Brown | @avakar: in such a case a crash dump maybe the piece of information to be given to the code author (if there is something available and it really helps to reproduce the scenario - which is not always the case). This is in no way opposed to my answer. | |
| Jun 10, 2013 at 15:52 | comment | added | avakar | (That doesn't mean the reporter shouldn't try to be helpful and provide the steps if they know them. For sporadic crashes, however, the reporter has no obligation to burn time on researching something the component owner will probably figure out faster.) | |
| Jun 10, 2013 at 15:49 | comment | added | avakar | The reporter shouldn't have to provide reproduction steps; quite often we simply attach a dump from the crashed process, especially if it occurs during an automated run. It is the assignee's responsibility to find reproduction steps so that the fix can be verified. | |
| Jun 10, 2013 at 15:27 | comment | added | Doc Brown | @user626528: IMHO the library owner is ought to try to go through the steps you tell him to reproduce the bug - he is not ought to try 500 slightly different scenarios when your description does not show up any bug. | |
| Jun 10, 2013 at 15:23 | comment | added | Doc Brown | @Heinzi: if I get a bug report "app crashes every now and then", I would give that issue a very low priority to work on, too. The minimum thing I expect the user is to write down how often is "now and then", what he was doing exactly what with the application at the time the app crashed the last time, and the exact error message. | |
| Jun 10, 2013 at 15:15 | comment | added | Heinzi | So if an app crashes "every now and then", without dicernible pattern for the user, then the developer does not have to fix it because the user cannot reproduce it on command? I strongly disagree here... | |
| Jun 10, 2013 at 12:54 | comment | added | user626528 | From the perspective of "standard workflow", does this mean that the library owner is not ought to try reproduce bug on his side? | |
| Jun 10, 2013 at 12:07 | history | edited | Doc Brown | CC BY-SA 3.0 | added 4 characters in body |
| Jun 10, 2013 at 12:01 | history | answered | Doc Brown | CC BY-SA 3.0 |