|
From: Yeshurun, M. <mei...@in...> - 2006-03-29 13:25:44
|
Hi, Can using --error-limit=3Dno (instead of suppressions) cause a = significant performance degradation, and if so, how much? Thanks, Meir |
|
From: Julian S. <js...@ac...> - 2006-03-29 13:51:54
|
> Can using --error-limit=no (instead of suppressions) cause a significant > performance degradation, and if so, how much? Yes -- that's why the limit exists in the first place. As you know V only displays each error the first time it appears, and so if your program has huge numbers of errors it can spend a lot of time checking incoming errors to find out if it has seen them before already. Note, the cost of this search is proportional to the number of different error call-stacks involved. So if the same error is seen one million times that's no big deal, but if you have one million different errors each of which have only happened once then the cost of checking a new error will be huge. Does that make sense? J |