On Sat, Dec 28, 2019 at 09:25:18AM +0100, Takashi Iwai wrote:
Mark Brown wrote:
There's two bits there - one is that it's adding to the mail volume when people chase up, the other is that if things have been lost then almost always the answer is that I don't have the patch any more (or it'll be error prone to find) and it'll need a resend so it's better to chase up by resending the patch since that can be acted on directly.
Well, I see a few points to be revised in this policy:
- If it were actually your oversight, then resending the patch makes sense. But if it's not merged by other reasons? Silently resending a patch can be worse.
For example, suppose the submitter either overlooked or didn't receive a reply or a followup in the thread. You didn't apply the patch because of the reply/followup pointing some problem. Now, the submitter tries to resend the original patch again without asking the situation, just because you suggested so in the past. You'll get the problematic patch again, and there is a risk that the patch gets merged mistakenly at this time.
The thing there is that if I don't remember the state of the patch I'm likely to just say "send it again" and if I do remember I'll remember either way (the form does say stuff about addressing feedback, though obviously people can ignore that too).
- The mail archive (lore.kernel.org) nowadays catches all posted mails in a proper manner, and it's pretty easy to fetch. And, resending the whole patch would be even higher volume than replying, disconnecting the discussions in the previous thread.
That requires me to be on the internet and fire up my web browser! I do actually work offline routinely, when I'm travelling or somewhere like a coffee shop with poor internet access so that's not always a thing I can do. I do postpone things but that's usually for longer periods (waiting for other reviewers and so on) which tends to mean people don't get an answer for their ping promptly which doesn't help either, I haven't managed to come up with a workflow for that which is effective.
So, I find it's OK to give this kind of warning for educational purposes to the people who don't know the common practice and send the patches too frequently. But for other cases, such a warning doesn't fit.
I deliberately try to be consistent in sending this stuff out because I don't want to be unfair. Which has it's own downsides as you say.