<samueldr>
superherointj: you were able to verify this fixes your problem, right?
<superherointj>
Yes
<superherointj>
in ARM, it fixes libvirtd.
<samueldr>
I mean, you tested it, right?
<superherointj>
In x86, worked allright.
<superherointj>
Tested like how?
<superherointj>
It stopped the kernel panics after enabling libvirtd.
<samueldr>
well, running libvirtd on ARM with the edited Nixpkgs or an overlay :)
<samueldr>
right, then sounds like you did
<superherointj>
Yes, I have tested it a couple of times. Without patch it crashes. With patch, it doesn't crash in ARM.
<superherointj>
But I have included other recommended patches beyond the patch I needed too.
<samueldr>
yeah, just wanted to make sure you didn't just update the package build, and was waiting on a Nixpkgs update to test :)
<superherointj>
;)
cript0nauta has joined #nixos-aarch64
Darkmatter66 has joined #nixos-aarch64
<superherointj>
Mic92, got my first PR. :)
CyberManifest has quit [Quit: Leaving...]
<samueldr>
superherointj: the only thing I would have done differently, is to genericize the fetchpatch
<samueldr>
first, we don't *require* the name, so that could have been left out I think
<samueldr>
(or be generic)
<samueldr>
so something like let recommendedPatch = commit: sha256: fetchpatch { inherit sha256; url = "https://.../${commit}" };
<samueldr>
then used like (recommendedPatch "1d0db85949a5bdd96375f6131d393a11204302a6" "11s8jciw7xf2668v79qcq2c9w2gwvm3dkcik8dl9v74p654y1nr8")
<samueldr>
though, that's really only an implementation detail that would have given (almost) the same result
<samueldr>
I don't know that we have a preference, as a community of maintainers, on that kind of helper function
<samueldr>
I have taken a peek at the patches, they are all relatively benign, or laser-focused, so I'm confident they'll not cause issus
<samueldr>
issues*
Darkmatter66 has quit [Ping timeout: 258 seconds]
Darkmatter66 has joined #nixos-aarch64
<superherointj>
Backporting to 20.03 would be nice. Because if libvirtd is enabled and someone happens to revert to the stable channel (20.03) it crashes the system.
<samueldr>
downgrading between versions is kind of an unspoken issue
<samueldr>
e.g. 20.09 -> 20.03, there is no guarantees, though generally it's fine
<samueldr>
but yeah, I was also thinking about the simpler case of someone today trying to set that up!
<superherointj>
Oh I thought NixOS could switch between channels without issues.
<samueldr>
the system itself can... but stateful software might not like that
<samueldr>
e.g. nextcloud version X on 20.09, but version X-2 on stable, its database schema might not work anymore
<samueldr>
that's because of that one inconvenient truth: state
<samueldr>
if you don't have state, then it should be generally fine
h0m1 has quit [Ping timeout: 240 seconds]
h0m1 has joined #nixos-aarch64
Darkmatter66 has quit [Quit: ZNC 1.7.5 - https://znc.in]
<superherointj>
samueldr, thanks for all the help! I'd wouldn't have done it with it.
<samueldr>
n/p thanks for doing the changes, sometimes going from the concept or idea to the change is the hard part :)
<samueldr>
and hopefully you got something useful out of that
<superherointj>
*without
<superherointj>
I've learned a lot.
superherointj has quit [Quit: Leaving]
Darkmatter66 has joined #nixos-aarch64
justanotheruser has joined #nixos-aarch64
rajivr has joined #nixos-aarch64
knerten1 has joined #nixos-aarch64
knerten has quit [Ping timeout: 258 seconds]
orivej has joined #nixos-aarch64
cript0nauta has quit [Remote host closed the connection]
cript0nauta has joined #nixos-aarch64
cript0nauta has quit [Ping timeout: 264 seconds]
cript0nauta has joined #nixos-aarch64
orivej has quit [Ping timeout: 265 seconds]
orivej has joined #nixos-aarch64
veleiro has joined #nixos-aarch64
alp_ has joined #nixos-aarch64
alp_ has quit [Ping timeout: 272 seconds]
* colemickens
smashes f5 on mesa 20.2
juliosueiras has quit [Ping timeout: 258 seconds]
alp_ has joined #nixos-aarch64
zupo has joined #nixos-aarch64
<ashkitten>
colemickens: ooh, it's that day already?
stiell has quit [Ping timeout: 246 seconds]
<colemickens>
supposedly this week, so I'm hoping today/tomorrow!
<ashkitten>
oooo
<ashkitten>
ACO by default here we come!
<colemickens>
actually prompted me to figure out how to rss sub to some repo's github tags :)
cole-h has quit [Quit: Goodbye]
stiell has joined #nixos-aarch64
lopsided98 has quit [Quit: No Ping reply in 180 seconds.]
stiell has joined #nixos-aarch64
lopsided98 has joined #nixos-aarch64
stiell has quit [Ping timeout: 260 seconds]
stiell has joined #nixos-aarch64
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
justanotheruser has quit [Ping timeout: 246 seconds]
Thra11 has joined #nixos-aarch64
pbb has quit [Ping timeout: 246 seconds]
pbb has joined #nixos-aarch64
alp_ has quit [Remote host closed the connection]
alp_ has joined #nixos-aarch64
superherointj has joined #nixos-aarch64
<superherointj>
Regarding the backport, do I still have action to take?
<superherointj>
Should I also push same patches to branches "nixos-20.09" and "nixos-20.03"?
<superherointj>
samueldr, understood what you were saying. Your idea is very clever.
<superherointj>
But the checksum won't match.
<superherointj>
Savanna's ID is not equal to NixOS's sha256sum.
<superherointj>
It would have worked if they were equal. I've read in the manual Nix (imprecise use of name?) normalizes the file before doing sha256sum. I guess maybe they have created the ID from a non-normalized file. So there might be the difference.
<superherointj>
If I were to change the file, I'd have to redo all the manual tests I did, and as my RockPro64 crashes for real (on boot), I'd have to take SDCard out fix it in another computer, and try next test. While the code currently might be a bit verbose. I have tested in different computers already. (2 arm boards, 1 x86) I'd preffer to sticking as it is then. Also, when `dmidecode` release happens, all these recommended patches will be dropped. So
<superherointj>
it is mostly a temporary patch.
<superherointj>
I plan using PXE eventually, I just haven't reached this point.
<superherointj>
And I like having the patch name so it can aid in finding any specific patch.
alp_ has quit [Ping timeout: 272 seconds]
alp_ has joined #nixos-aarch64
zupo has joined #nixos-aarch64
orivej has quit [Ping timeout: 260 seconds]
zupo has quit [Ping timeout: 256 seconds]
zupo has joined #nixos-aarch64
<superherointj>
Is there some automated way to make backporting easier? I'd manually copy and paste the code to submit to the new branch. What method is more efficient for this task?
superherointj has quit [Quit: Leaving]
tilpner_ has joined #nixos-aarch64
tilpner has quit [Remote host closed the connection]
tilpner_ is now known as tilpner
W1lkins has quit [Quit: Bye]
evalexpr has joined #nixos-aarch64
evalexpr has quit [Client Quit]
evalexpr has joined #nixos-aarch64
<Mic92>
I am also look for a tool that can do it.
<simpson>
There's git-cherry-pick, but I don't think that that's what you're asking for.
<Mic92>
Yeah. a bit more automation than git-cherry-pick.
<Mic92>
something that could take a pull request number and the target branch would be nice.
juliosueiras has joined #nixos-aarch64
evalexpr has quit [Remote host closed the connection]
evalexpr has joined #nixos-aarch64
evalexpr has quit [Remote host closed the connection]
evalexpr has joined #nixos-aarch64
evalexpr has quit [Client Quit]
evalexpr has joined #nixos-aarch64
superherointj has joined #nixos-aarch64
alp_ has quit [Ping timeout: 272 seconds]
alp_ has joined #nixos-aarch64
orivej has joined #nixos-aarch64
alp_ has quit [Ping timeout: 246 seconds]
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
Acou_Bass has quit [Ping timeout: 260 seconds]
Acou_Bass has joined #nixos-aarch64
superherointj has quit [Quit: Leaving]
alp_ has joined #nixos-aarch64
<juliosueiras>
the serial cable is taking its sweet sweet time
<juliosueiras>
to get shipped
<juliosueiras>
(correct, is not even shipped)
<cirno1000>
julio, nixos on vita when ( ͡° ͜ʖ ͡°)
zupo has joined #nixos-aarch64
zupo has quit [Ping timeout: 240 seconds]
justanotheruser has joined #nixos-aarch64
veleiro has quit [Read error: Connection reset by peer]
cole-h has joined #nixos-aarch64
juliosueiras has quit [Ping timeout: 260 seconds]
superherointj has joined #nixos-aarch64
zupo has joined #nixos-aarch64
<superherointj>
Backporting to 20.09: https://github.com/NixOS/nixpkgs/pull/97686 ... I'm a bit unsure about what I am doing. I've just copied and pasted to 20.09. Will do the same to 20.03.
<samueldr>
this chapter explains that we use `git cherry-pick -x` to pick the commit into stable branches
<superherointj>
Oh. I had no idea about this.
<samueldr>
understandable, if no one told you :)
<samueldr>
I added the label that it needs port to stable more for other committers than for you
<samueldr>
the basic idea is that with -x git can do a bit more internal bookkeeping about the commit... but more importantly, it shows other contributors on stable branch where the change comes from
<superherointj>
Should I cancel the pull request?
<samueldr>
you can re-do it and force push to the branch, it's fine too if you want to do the work :)
<samueldr>
though, remember I asked about git and squashing?
Darkmatter66_ has joined #nixos-aarch64
Darkmatter66 has quit [Ping timeout: 264 seconds]
<samueldr>
it'd help (1) to pollute the heavy git history less with "fixup" commits that have no "semantic" meaning, and (2) to facilitate your backport by cherry-picking one commit
stiell has quit [Ping timeout: 264 seconds]
<samueldr>
ideally each commit is a logical meaningful change in Nixpkgs, ideally focused into a particular requirement or task
<superherointj>
So, I need to (1) squash (2) cherry-pick. Then it will be fine without having to cancel the PR. Right?
stiell has joined #nixos-aarch64
<samueldr>
yes
<samueldr>
while I wasn't a git noob when I started using and contributing to Nixpkgs, it sure helped me with bite-sized tasks in learning more git
rajivr has quit [Quit: Connection closed for inactivity]
justanotheruser has quit [Ping timeout: 240 seconds]
<superherointj>
samueldr, the squash worked but when I got to try to reverse the `git commit` I had previously done to do the cherry pick, well, things got weird.
<samueldr>
weird how?
<superherointj>
Problem is I have edited other branches. And I was trying to reverse the changes. So I "git rebase -i ###^" to previous commit. And force pushed.
<superherointj>
Now it is asking for review of: Mic92 and zowoq as code owners
<samueldr>
right
<samueldr>
I see
<samueldr>
seems you "edited" the commit for the github action thingy
<superherointj>
How should I have reversed the commit?
<samueldr>
there are multiple strategies
<samueldr>
first off I see you didn't create a "topic branch"
<samueldr>
it really should be done when you work on something so you can have concurrent changes
<samueldr>
git checkout -B my-topic-branch origin/nixos-20.09 # where you probably want a name like 20.09/dmidecode-recommended-patches rather than my-topic-branch
<samueldr>
github is somewhat annoying; I don't think you can change the source branch of a PR
<samueldr>
what you could do though, is continue with using the nixos-20.09 branch name for now
<samueldr>
oh, be careful, I used -B, I should have suggested -b
<samueldr>
-B will overwrite an existing local branch name
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<samueldr>
git checkout -B nixos-20.09 origin/nixos-20.09 # will "create a new branch" or "set" the branch name to the commit represented by origin/nixos-20.09
<samueldr>
forcibly
<samueldr>
then you could cherry-pick -x and force push
zupo has joined #nixos-aarch64
<superherointj>
Is there some way I could drop my changes? So I can start clean instead of getting the mess right?
<samueldr>
(what I dislike with git is how many ways you can do the same thing, that are all semantically different enough to matter)
<samueldr>
that's what I described
<superherointj>
Right.
<superherointj>
But I've got the PR opened in GitHub and that is tied to the branch name.
<superherointj>
So I just close the PR?
<samueldr>
git checkout -B $local_branch_name $commitish # will forcibly reset and checkout at the state of the parameter
<samueldr>
you could continue using the nixos-20.09 branch name if you don't care that much
<superherointj>
Then I git push --force? To have it cleaned remotelly?
<samueldr>
you should push --force (or --force-with-lease for good practice) only after you also have the cherry-pick
<samueldr>
otherwise github will close the PR since there's no changes
<samueldr>
another tip, when opening a backport PR, prefix the PR title with the version, e.g. [20.09]: or 20.09:
alp_ is now known as alp
justanotheruser has joined #nixos-aarch64
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
zupo has joined #nixos-aarch64
zupo has quit [Client Quit]
alp has quit [Ping timeout: 272 seconds]
<superherointj>
It seems the cherry pick worked. Much easier!
<superherointj>
(This link above was the backport for 20.09)
zupo has joined #nixos-aarch64
* superherointj
wonders if has caused any collateral damage by doing this.
<samueldr>
nothing worse than pinging those committers, until it's merged
<samueldr>
I wonder why you still have that commit in your history for the nixos-20.09 branch
<samueldr>
you could rebase it out, by rebasing against origin/nixos-20.09
<samueldr>
I'm sure there are at least 5 other ways to do that with git
zupo has quit [Ping timeout: 258 seconds]
<superherointj>
I think I'm still doing something wrong in GIT. What I did: (1) git log -n2 (2) find previous commit hash (3) git rebase -i ####^ (4) The previous commit does not exist in 'git log -n2'. Previous commit is a 'cherry pick', can it influence?
Darkmatter66 has joined #nixos-aarch64
zupo has joined #nixos-aarch64
Darkmatter66_ has quit [Ping timeout: 256 seconds]
<samueldr>
I would rebase on the name of the branch rather than trying to find your way around using those magic cursed git symbols
<superherointj>
In the rebase editor opens w/ 3 lines of commits. But as I've picked the previous commit. I was expecting only two lines. It seems that is how I am messing up with someone else's code.
<samueldr>
git rebase --interactive origin/nixos-20.09
<superherointj>
My remote repository is my fork, not official nixos. When I do this rebase I end up redownloading my patches.
<superherointj>
Oh
<superherointj>
Maybe I should use a different remote!
<samueldr>
oh, right, duh, I assumed that you had cloned from the nixos repo, and origin was it
<superherointj>
Good.
<samueldr>
obviously instead of origin/nixos-20.09 it would be whateverremotename/nixos-20.09
<Dandellion>
`nix build nixpkgs.pkgsCross.whatever.nur.repos.foo.bar` not working was a little dissapointing :P
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
juliosueiras has quit [Ping timeout: 256 seconds]
alp has joined #nixos-aarch64
cript0nauta has quit [Remote host closed the connection]
cript0nauta has joined #nixos-aarch64
t184256 has left #nixos-aarch64 ["Disconnected: Replaced by new connection"]
t184256 has joined #nixos-aarch64
juliosueiras has joined #nixos-aarch64
ryantrinkle has quit [Ping timeout: 258 seconds]
bennofs_ has quit [Quit: No Ping reply in 180 seconds.]