Hi Tom, Andreas,
On 7 August 2016 at 14:07, Tom Rini trini@konsulko.com wrote:
On Sun, Aug 07, 2016 at 06:46:37PM +0200, Andreas Färber wrote:
Am 07.08.2016 um 15:31 schrieb Tom Rini:
On Sat, Aug 06, 2016 at 06:05:29PM +0200, Andreas Färber wrote:
Hi Simon,
Am 06.08.2016 um 06:30 schrieb Simon Glass:
Hi Andreas,
On 17 July 2016 at 19:06, Andreas Färber afaerber@suse.de wrote:
Hi,
This series adds initial support for RK3368 SoC and GeekBox. For more details see the commit message.
Will need to be rebased onto Heiko's cleanups and Kever's RK3399 series.
Regards, Andreas
Cc: Simon Glass sjg@chromium.org Cc: Kever Yang kever.yang@rock-chips.com Cc: Heiko Stübner heiko@sntech.de
Andreas Färber (2): dts: Import rk3368-geekbox.dts ARM64: rockchip: Add initial support for RK3368 based GeekBox
Are you planning to respin these patches?
Eventually...?
I'd like to get them applied soon.
And I'd like to get my work recognized! However, despite our previous IRC chat, I had to find out _while_ replying to the rk3399 mails that you had once again not just applied all patches (twenty minutes after ack'ing them on a Saturday) but already sent a pull on Tuesday my nighttime that I was not CC'ed on and that Tom has merged the night after. So it feels like I'm wasting my time here and consequently I stopped my review and rebase.
In the U-Boot community, we are not in the habit of cc'ing everyone with a change in a given pull request. Is there a tool the kernel folks use here that makes this easy?
Not that I'm aware of.
But that is besides the point, as my very complaint is that I'm not credited in the patches that got merged, so no tool could've extracted my name for CC'ing.
Alright.
It's about Simon having mismerged those patches and having overlooked unresolved review comments of mine for those patches before and me specifically having complained to him about not waiting for my Reviewed-by before applying them. So him seeing that I did not reply to his Saturday mails, I feel it would've been fair in this particular case a) to ping me again after the weekend and b) to let me know that he is no longer waiting for my review comments or that I really need to hurry up with an objection until X. He did not say so in a reply that reached my inbox, and I was not CC'ed on the pull request itself, thus a pull request behind my back.
I'm not too deep into U-Boot, so maybe there was a reason for this hush-hush workflow, but then at least the communication was fairly bad.
Had I known that the pull is already on the list, I wouldn't have replied with a Reviewed-by for 1/4 that same day (which surely Simon was CC'ed on) or I could've asked Tom to hold off merging it until I'm done reviewing the next day.
And the rule of thumb that I use, and I try and get everyone else to use as well is that a patch should be out for a week before it gets picked up and merged as that should give everyone time to review, comment and test. Did that not happen with the patches Simon picked up?
Slightly less than a week. For some other projects it's ~two weeks. Also again note that this is not about some random patch but one where Simon specifically said he would be away, that he would exchange the patches on his branch where necessary and where he asked me to "sing". It leaves a bad taste that Simon was absent himself the week the patches were posted but apparently expects me to be available whenever he is. I don't work on U-Boot as a job, and for rebasing rk3368 patches - which many of my review comments resulted from - I need access to the hardware for testing.
Note that I was similarly surprised how quickly two patches of mine went into his tree, with just one day in between and despite conflicts between my rk3368 and Kever's rk3399 preparations. I can see that having patches in a tree facilitates testing, but it also prevents serious peer review when not just staging but also merging them.
I want to treat all of the above at once. First, sorry. We don't have an intentionally "hush-hush" workflow, but every custodian does decide how many emails to send when moving a patch forward. And unless I'm testing multiple PRs (or they come in while I'm already testing one) the time between getting a PR and applying it is usually the same (US, east coast) day if it passes my testing. But we are trying to include more crediting, not less, so it is not intentional to have left things out[1]. So this was a mistake in our part, sorry. Sometimes review comments are missed. But this too is not usually intentional unless it's small things that can be addressed in a follow-up in order to get things otherwise in and unblocking other work. In the end however, Simon, please slow down a bit. It's OK if stuff misses a release, there's always the next one. And it may make sense to follow what Hans has been doing with the sunxi tree and having a next branch (which can always be rebased and stuff replaced in!) so it's both clear that patches are being picked up, but not rushing to get things in before a window closes. Thanks!
OK thanks Tom for looking at this. I'll slow down a bit on the Rockchip stuff. I'll look at using a -next branch for that also.
Andreas let me know how things go. I'd really like to see a review for new patches within a week if possible (absent holidays, etc.).
-- Tom
[1]: Patchwork had an outage recently and some stuff didn't make it in there and thus was missed unless picked up manually. Most tags come in via patchwork not manually adding to the commit message.
Yes that was a pain!
Regards, SImon