Messages in this thread | | | Date | Thu, 5 Feb 2015 17:07:43 -0500 | From | Steven Rostedt <> | Subject | Re: [RFC][PATCH 5/7] trace: make filter_parse_regex() provide the length of substring to compare with |
| |
On Thu, 5 Feb 2015 21:44:24 +0000 Al Viro <viro@ZenIV.linux.org.uk> wrote:
> The point is that by now this strlen() is the only thing for which we > NUL-termination of the substring; moving it inside the filter_parse_regex() > is an obviously equivalent transformation and it leaves that one strlen() call > inside filter_parse_regex() the only place where we still care about NUL. > > The next commit kills it off completely, at which point we are done with > modifying the string at all.
Thanks for the explanation,
Can you add that to the change log. I like to think patches can stand on their own, and if they are added to help another patch, it should be stated in the change log so someone doing a git blame followed by a git show, knows what is going on.
Changes done for a patch series may be fine when that series is being reviewed, but a change log is the only thing we have in the future to understand why something was done. And a change done to help out another change is lost when going back and looking at the history.
I've been burnt by my own code by looking back at why I did something and not having a change log on a change describing things to why they were done (because at the time it seemed obvious). But then I wasted a lot of time digging through archives to figure out the history of some change. I rather avoid doing that again.
Thanks,
-- Steve
| |