Re: [Nix-dev] Fwd: Why having releases if you break things in it often

[Vladimír Čunát]

On 02/12/2017 01:57 PM, Stefan Huchler wrote:
> So is there a way to rollback the --update action, so that new
> configurations still use the old packages? Else for my problem
> --rollback is no solution.

Yes, channel updates can be rolled back simply by nix-channel --rollback.




smime.p7s
Description: S/MIME Cryptographic Signature
___
nix-dev mailing list
nix-dev@lists.science.uu.nl
http://lists.science.uu.nl/mailman/listinfo/nix-dev

Re: [Nix-dev] Fwd: Why having releases if you break things in it often

[Stefan Huchler]

Vladimír Čunát  writes:

> On 02/11/2017 03:26 PM, Stefan Huchler wrote:
>> Maybe we could talk about the status of the project wiki or other
>> documentation, if there would ways to improve that.
>
> Status of the wiki is being discussed right now:
> https://github.com/NixOS/nixpkgs/issues/22599 and the rest of
> documentation is also touched lightly in there.
>
>
> I'm convinced the stable releases are worth it!  In the last several
> months NixOS has been patching every single CVE reported in a larger
> distribution (through LWN tracker) and backporting all those to the
> stable releases.  I really find it *amazing* that we manage that
> consistently.  There are occasional breakages due to some of the
> updates, but that's hard to avoid completely and I don't think there
> have been many.  (There's also the advantage of --rollback :)

I did not claim that stable releases are not worth, that was the
original suspect from 1 2 months ago, and it was just a question because
I wanted to learn what policies you have.

To the rollback feature, the problem is that you have no real seperation
between configuration and packaging.

That leads for me to have 1 packaging set that works ( a old version
from last year), but with a wrong vpn server write-only in the openvpn
settings) and a new profile with the correct openvpn configuration but a
broken openvpn package version.

So is there a way to rollback the --update action, so that new
configurations still use the old packages? Else for my problem
--rollback is no solution.

___
nix-dev mailing list
nix-dev@lists.science.uu.nl
http://lists.science.uu.nl/mailman/listinfo/nix-dev

Re: [Nix-dev] Fwd: Why having releases if you break things in it often

[Vladimír Čunát]

On 02/11/2017 03:26 PM, Stefan Huchler wrote:
> Maybe we could talk about the status of the project wiki or other
> documentation, if there would ways to improve that.

Status of the wiki is being discussed right now:
https://github.com/NixOS/nixpkgs/issues/22599 and the rest of
documentation is also touched lightly in there.


I'm convinced the stable releases are worth it!  In the last several
months NixOS has been patching every single CVE reported in a larger
distribution (through LWN tracker) and backporting all those to the
stable releases.  I really find it *amazing* that we manage that
consistently.  There are occasional breakages due to some of the
updates, but that's hard to avoid completely and I don't think there
have been many.  (There's also the advantage of --rollback :)


--Vladimir




smime.p7s
Description: S/MIME Cryptographic Signature
___
nix-dev mailing list
nix-dev@lists.science.uu.nl
http://lists.science.uu.nl/mailman/listinfo/nix-dev

Re: [Nix-dev] Fwd: Why having releases if you break things in it often

[Stefan Huchler]

>
> Did you pay someone?  If yes, talk to them and be as rude as you want,
> because you have a contract.  If no, then pick one of the other three
> options.

Well I wanted to appoledgise for my tone, even I think I made mostly
valid points if you put the tone aside for a moment.

I of course have no RIGHT to get help, but I think I never said
that. But many of you get really triggered, even I tried to give even my
rant messages a possitive spin at the end.

Yes I could been more polite, everybody has a bad day here and there, I
think I did not overstep any real nogos, like attacking a person or
getting personal or anything.

And if I critisize you and I dont know how many of you listen here 100
people? and 90% then say I will not help you if you are not more polite,
I have to live with that. But maybe the other 10 people can put their
ego aside for a moment and try to fix it.

Its not like a bug report is only a onesided thing, where only the
person that writes the bugreport gets somethnig out of it. Its a very
important work people do to HELP the project.

They might not do that because of that reason to help the project, but I
am really happy if people send me bug reports to my software.

And I am really happy if people use my software. Yes its better to be
polite 100% of the time, but sometimes a rant can help.

Linus did show Nvidia the middle finger for bad linux support. They kind
of stepped up their game afterwards.

Btw I dont compare you to nvidia, but lets not start to talk about
nvidia to much :)

You could argue that nvidia customers give nvidia money, so they could
expect good driver support even for linux, or at least more than this
"freeloader" that I am.

Again I release 100% of my software under the gpl, I posted here a link
to a derivation, I wrote a free kodi emacs remote which is in melpa.

So its not like I am a asocial person that only wants stuff but never
gives stuff back.

I just feel my time is better invested maybe add a feature to
kodi-remote as example than in repackaging a broken package, and
learning the complete nix language and api / structure.

And I am willing to learn, on the way, if somebody would send me a fix
or workaround I would learn from it. But without a good documentation
(imho), and a very complex system its hard to do on your own.

And critisise me for my rant (if you do that every time something is not
working out the way you want or not fast enough I get that you dont want
that, but I dont rant all the time, about technical stuff, I cant
remember the last rant at least on a mailing list), but it kind of
worked out, on a related bug there was more movement in the last 12
hours then is the last 7 days.

there seems to be work to fix it going on and another attempt to give me
a workaround. Which I have to test out today.

Would that happend also if I asked more politly maybe, I did not do that
calculated in how much affect it may have, I just needed a ventil for my
adrenalin or testestoron I guess.

Again sorry for the tone, but it can be very frustrating if you try to
go through several doors but instead always run against a wall.

Maybe we could talk about the status of the project wiki or other
documentation, if there would ways to improve that.

Its just hard to come up with better docu if you dont understand most
stuff on yourself.

At least I will send more bug reports when something does not work and
hope you see that as a contribution and not a burden.

And I am of course thankful for any help, and hope that the workaround
@bjornfor gave me on github solves the issue.

See ya!

___
nix-dev mailing list
nix-dev@lists.science.uu.nl
http://lists.science.uu.nl/mailman/listinfo/nix-dev

Re: [Nix-dev] Fwd: Why having releases if you break things in it often

[Ertugrul Söylemez]

>>> I dont want to be a 100% full time nixos developer / sysadmin just
>>> to solve such stupid problem. Except somebody would pay me for that,
>>> then I would think about it.
>>
>> Ah, so you won't do it for free but you expect the army of experts to
>> rush for help and do it for free for you, do I read that correctly?
>
> "rush" the bug is open since over a month, I would not call that
> "rush".

That's not the point, is it?  NixOS is not a commercial organisation.
It's entirely driven by voluntary work.  Yes, even the companies that
contribute to NixOS are volunteers!  In other words you can do one of
four things:

  * Contribute patches yourself.

  * Kindly (!) ask for help and with a bit of luck receive voluntary (!)
support from the community.  Remember that nobody is obliged to help
you, but if you're nice and respectful, people will usually try
their best in my experience.

  * Use a different solution.

  * Oblige somebody to help you by paying them.

Did you pay someone?  If yes, talk to them and be as rude as you want,
because you have a contract.  If no, then pick one of the other three
options.


> If thats only a very unproffessional project, why creating a
> professional looking web site?

Why not?  We are not obliged to *anything*, not even a terrible
website. =)


> So you find time for much PR, but you cant answer a bug report in a
> month?

Sure, why not?  Who are you to tell me what I should do in my spare
time?  Pick one of the four options above.


Greets
ertes


signature.asc
Description: PGP signature
___
nix-dev mailing list
nix-dev@lists.science.uu.nl
http://lists.science.uu.nl/mailman/listinfo/nix-dev

Re: [Nix-dev] Fwd: Why having releases if you break things in it often

[Stefan Huchler]

Tomasz Czyż  writes:

> But you know, experts stuff is pretty expensive resource, and maybe
> people decided will be better if they use it differently than you
> would expect?

So you find the time to create Regressions, but none to fix the Mess you
created?

Creating Regressions, is basicly the worst a developer can do, if he
would have done nothing it would be still better than creating a Regression.


>  I dont want to be a 100% full time nixos developer / sysadmin just to
>  solve such stupid problem. Except somebody would pay me for that, then I
>  would think about it.
>
> Ah, so you won't do it for free but you expect the army of experts to
> rush for help and do it for free for you, do I read that correctly?

"rush" the bug is open since over a month, I would not call that
"rush". 


> And now serious, nix is an effort of huge number of people, they
> contribute in their free time mostly, and even those who do it for
> cash, mostly work for some companies and nixos contribution is just
> side effect of their work.

If thats only a very unproffessional project, why creating a
professional looking web site?

So you find time for much PR, but you cant answer a bug report in a month?

> It looks like a lot of people think that
> nixos is so useful for their work they contribute to it and even help
> others. You can join to that movement and spend a lot of time learning
> about this open source project or you can use the company to help
> you. Few people started nixos consulting (me as well :-) and this is a
> second option, you pay for product/service. It's like most of the open
> source software, right?

Most other opensource software have good documentation, nixos primary
documentation is the sourcecode of the package-files or service files.

Also most opensource projects answer stuff on mailing lists, here I only
got told I should use irc or github bug tracker. Which is a pain in the
ass, cause I dont use irc and freenode has some restrictions and I had
in the past somebody allow my ip adress or somethnig because else it
would not work.

Also I hate browser interfaces and prefer mailinglists, but despite
that, I wrote a bug report there and got no answer in a month. Which btw
is also not standart for other opensource projects, only basicly dead 1
person projects dont answer for more than a week.

But instead of trying to discuss further if I am right or not, could
somebody just help me despite our maybe disagreing oppinion what I
should accept.

I wrote a derivation for another package I think nobody cared about
it. I dont care to much, because its not a central thing for me. So I am
not only complaining about bugs, but if I have time and its not
something from my productive "critical" systems, I dont care to invest
here and there some time.

But if something that worked fine just breaks from a stable update, I am
not willing to invest 20-50 hours to solve it, cause somebody else
did something wrong.

You can critisize my tone, but I suggest you stand over that and try to
step up your game a bit, and I am not talking about you personaly.

Again I dont even want a REAL solution, just a quick dirty hack to work
around that 2 3 nixos commands with the 50 options every command has,
that for you seem to be very simple and for me is rocket science with
very bad docu on it.

Of course you cant expept anything but death from live, but you dont
have to take me so literly. I just expect it, cause its what people get
in other mailinglists. Same basic help.

Real professionals take some angree words not that personal, if they are
not very disrespectful and under the belly.

The frustration is just not because I tried one thing, and it did not
work and I immidietly bitched about it.

Its just if you try several different things and nothing works:

1. mailing list no useful answer
2. bugreport no answer
3. doku horrible
4. tried some stuff that seemed to related
5. read manpages

and you nowhere see any lever to get it fixed, after 1 month or so, I
get impatiant.

I dont want somethnig fancy, I dont even have a feature-request, I just
want no regressions, or a fix for a regression, in a reasonable time.

I dont think thats that much to ask for.

___
nix-dev mailing list
nix-dev@lists.science.uu.nl
http://lists.science.uu.nl/mailman/listinfo/nix-dev

[Nix-dev] Fwd: Why having releases if you break things in it often

[Tomasz Czyż]

I'm not sure if that's a troll or not, but let's try:

So any help would be nice. I am a technicaly shurly keen guy, but I cant
> reboot daily my machine for months, cause nobody of you experts can
> invest 5-15 mins to fix that.
>
But you know, experts stuff is pretty expensive resource, and maybe people
decided will be better if they use it differently than you would expect?

>
>
> I dont want to be a 100% full time nixos developer / sysadmin just to
> solve such stupid problem. Except somebody would pay me for that, then I
> would think about it.
>
Ah, so you won't do it for free but you expect the army of experts to rush
for help and do it for free for you, do I read that correctly?

>
> Is that really what I have to expect?
>
:-)

>
> Sorry for ranting here, I stopp now, but maybe some of that feedback
> helps you to improve something, and maybe somebody could give me some
> config lines or nix-env commands that gets flexget running again.
>
I can improve, but you know... if you would pay me, than I would think
about it.
Sorry, just kidding :-)

And now serious,
nix is an effort of huge number of people, they contribute in their free
time mostly, and even those who do it for cash, mostly work for some
companies and nixos contribution is just side effect of their work. It
looks like a lot of people think that nixos is so useful for their work
they contribute to it and even help others. You can join to that movement
and spend a lot of time learning about this open source project or you can
use the company to help you. Few people started nixos consulting (me as
well :-) and this is a second option, you pay for product/service. It's
like most of the open source software, right?


> Graham Christensen  writes:
>
> > I'm very sorry you've had a bad experience with breakage on stable. :( I
> > use 16.09 myself.
> >
> >>> yes I think that html5lib thing would it be. So it was at least a
> >>> security fix, so you dont just update stuff to update it, which would
> >>> make releases pretty useless concept :)
> >
> > Roughly, this is why backports happen:
> >
> >  - Security patches which aren't major updates
> >  - If a security patch is a major upgrade, try and find patches to our
> >current version which accomplish the same goal. Apply the major
> >update to master, and the patches to stable.
> >  - Bug fixes to applications which, again, aren't major updates.
> >Generally be cautious about these.
> >  - Any updates when the current stable version is utterly broken. A key
> >example of this is Spotify, who regularly breaks their old versions.
> >  - Extremely security-sensitive software, in particular Chrome,
> >Chromium, Firefox, Thunderbird, and of course the kernel.
> >
> >>> Sorry I formulated that message a bit trollish, but just wanted to
> learn
> >>> why how releases are done in nixos.
> >
> > Please know that Freddy, Franz, Robin, Domen, myself, and the rest of
> > the people contributing to NixOS work very hard to keep the stable
> > version of NixOS working nicely. This is very important to us.
> >
> > It can be very stressful when preparing to backport changes, but it is
> > important to do them anyway. I try to think through impact and run tests
> > across a wide range of software to see what will break. We also try not
> > to backport any substantial changes, but instead smaller patches to
> > prevent breakage.
> >
> > When you do find breakage, please do promptly open an issue on send a
> > report on the mailing list so we can address the problem and perhaps add
> > testing to prevent it in the future. We're also quite accessible on the
> > #nixos IRC channel on Freenode.
> >
> > If you would like to take part in the process of identifying and solving
> > security problems on master and backporting to stable, we sure would
> > love the extra help -- feel free to comment on
> > https://github.com/NixOS/nixpkgs/issues/21967 and I'll tag you on
> > Wednesday when I open the next roundup.
> >
> > Best,
> > Graham Christensen
>
> ___
> nix-dev mailing list
> nix-dev@lists.science.uu.nl
> http://lists.science.uu.nl/mailman/listinfo/nix-dev
>



-- 
Tomasz Czyż



-- 
Tomasz Czyż
___
nix-dev mailing list
nix-dev@lists.science.uu.nl
http://lists.science.uu.nl/mailman/listinfo/nix-dev