Re: [gentoo-dev] Re: Ebuild Janitor Project -- Any formalities tojoining it?

Re: [gentoo-dev] Re: Ebuild Janitor Project -- Any formalities tojoining it?

[Joshua Brindle]

>Frankly people like you kill every possible joy I might get out of this
>- all you do is complain, not one word about the relative niceness of
>this project and what it could do for the state of the ebuilds.

Quite the contrary, I talked on irc about how the project could help
though i didn't mention it here, sorry about that.

>That being said - No, I didn't talk to anyone about it before going
>ahead, portage is gpl'ed, I can do what ever I want to it, and I intend
>to. I didn't want to announce anything untill it was off the ground
>because I think people spend FAR to much time talking and FAR to little
>time actually doing something. 
>I didn't expect it to take off this fast, we have been going for a few
>days, and there are now several ebuilds waiting in queue to be comfirmed
>working once that has happened they will be forwarded to bugzilla. (This
>all written in the guidelines on the forum, but I'm guessing you didn't
>even take your time to read that).

Yea, you can do whatever you want with it, I never said you could't.
The project has good goals and the ability to help, my ENTIRE question
which you did NOT answer was of logistics. Ebuild XYZ  sucks and janitor
fixes it and submits it to you, what then? If it doesn't somehow get into
our main portage tree than it can't help anyone. THIS is the reason
you should be talking to devs, to see if we can come up with a way 
for the cleaned up ebuilds to make it into the tree. 

>Now suggest you do some research before giving nice people like my
>teammates and myself a hard time.

what research? It doesn't matter how nice you and your teammates 
are if the project has no ability to deliver the product. Keep in mind
I never questioned you or your ability or your projects or your teammates.
I asked how you planned on getting this to work and you reacted
in a very emotional and defensive way. I suggest you evaluate the 
goals of the project and your ability to contribute to Gentoo with it.
We will be more than happy to facilitate your project and it's contributions
but you must take the initiative to talk to developers who can help you.


Joshua Brindle

--
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] Re: Ebuild Janitor Project -- Any formalities tojoining it?

[Marius Mauch]

On 11 May 2003 14:31:00 -0500 Joshua Brindle wrote:

> >That being said - No, I didn't talk to anyone about it before going
> >ahead, portage is gpl'ed, I can do what ever I want to it, and I
> >intend to. I didn't want to announce anything untill it was off the
> >ground because I think people spend FAR to much time talking and FAR
> >to little time actually doing something. 
> >I didn't expect it to take off this fast, we have been going for a
> >few days, and there are now several ebuilds waiting in queue to be
> >comfirmed working once that has happened they will be forwarded to
> >bugzilla. (This all written in the guidelines on the forum, but I'm
> >guessing you didn't even take your time to read that).
> 
> Yea, you can do whatever you want with it, I never said you could't.
> The project has good goals and the ability to help, my ENTIRE question
> which you did NOT answer was of logistics. Ebuild XYZ  sucks and
> janitor fixes it and submits it to you, what then? If it doesn't
> somehow get into our main portage tree than it can't help anyone. THIS
> is the reason you should be talking to devs, to see if we can come up
> with a way for the cleaned up ebuilds to make it into the tree. 

What is the recommended way then to get in touch with the devs? We
thought that sending a mail to -dev would be the best way. See
Lovechilds mail as a RFC. Your mail had some good points, but I think it
sounds quite harsh.

> >Now suggest you do some research before giving nice people like my
> >teammates and myself a hard time.
> 
> what research? It doesn't matter how nice you and your teammates 
> are if the project has no ability to deliver the product. Keep in mind
> I never questioned you or your ability or your projects or your
> teammates. I asked how you planned on getting this to work and you
> reacted in a very emotional and defensive way. I suggest you evaluate
> the goals of the project and your ability to contribute to Gentoo with
> it. We will be more than happy to facilitate your project and it's
> contributions but you must take the initiative to talk to developers
> who can help you.

Ok, we all calm down a bit. I can't talk for the others, but I see the
project as a testbed for ebuild modifications that can be tested by
users before they get submitted to bugs.g.o. As Lovechild has said there
are also discussions about new use flags and other modifications to the
current portage system. I missed the discussion on IRC but I saw that
you were backed by other devs. Please notice that this projects has been
started just for a few days so of course we haven't clarified all
issues.

Marius (aka Genone)

--
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] Re: Ebuild Janitor Project

[David Nielsen]

> what research? It doesn't matter how nice you and your teammates 
> are if the project has no ability to deliver the product. Keep in mind
> I never questioned you or your ability or your projects or your teammates.
> I asked how you planned on getting this to work and you reacted
> in a very emotional and defensive way. I suggest you evaluate the 
> goals of the project and your ability to contribute to Gentoo with it.
> We will be more than happy to facilitate your project and it's contributions
> but you must take the initiative to talk to developers who can help you.
> 

Okay, our guidelines clearly state that when a fixed ebuild has been
confirmed by two other people as working, we submit it to bugzilla. All
the ebuild janitors do is make sure that the ebuilds use the USE flags
to the maximum of their potential, and of course generally clean them up
as there's a lot of stuff in some older ebuilds that is no longer valid.
Your mail clearly showed that you had not read those guidelines..

The plan is NOT to fork portage, the plan is NOT to stage a coup, the
plan is NOT to start a pissing contest with the developers... the plan
is quite simple, to make portage even better.

How to make it work, it's simple, fix ebuilds, make sure they work,
submit to bugzilla.

Should we be so unlucky as to hit a major problem that we can't handle
within the project, we will of course take contact to the person in
charge, or mail gentoo-dev for input depending on the situation.

We have already hit some cases that could be viewed as imperfections in
the way flags are handled in Portage, and we have yet to receive any
technical feedback on these issues.

- Lovechild


--
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] Re: Ebuild Janitor Project

[George Shapovalov]

Hey guys.

As I was reading through last few day's postings I realised, there is certain 
important part missing in the picture of streamlined/distributed ebuild 
processing system mentioned earlier. That is of the place where users could 
actively interact and discuss the ebuilds thay submitted and get the fixes 
rolling as discussion brings our shortcomings... Looks like this project 
nicely fits to this role ;).
Well, there is this:
http://forums.gentoo.org/viewforum.php?f=8
so I kept forums at the back of my mind, but  looks like the forums you setup 
are trying to keep more to the point. So this is not really a duplication 
after all...

Now, on the integration issue, which is quite essential, as was already 
mentioned. The way I see it, distributed ebuild processing requires:
1. some central depository with automated submission system doing at least 
some basic checks (multiple access/stability levels are already in portage, 
but that might need some further ratification)
2. strong voting/feedback system (think gentoo-stable/stats), looped back to 
ebuild level adjutment (some automatic some manual).
3.(new!) the way for users to be getting feedback on their ebuilds and be able 
to discuss related stuf with each other. Apparently best done in user-land, 
by the project such as the one you initiated. Should really do a lot in terms 
of increasing overall quality of submitted ebuilds.

Ok, this by no means should be considered as detailed to any level or even 
setting any course of action, just trying to throw the overall structure into 
the mix, so that it gets registered and remembered later when need arises :), 
especially since there was a question where this project could fit ;). 
For those who did not follow related discussion few weeks ago, some more 
details on pp 1 and 2 are available in #1523, however outdated by all the 
changes that happened since and pending redesign which should follow soon 
after the internal reorganisation (mentioned in recent GWN) completion.

George


On Sunday 11 May 2003 14:01, David Nielsen wrote:
> Okay, our guidelines clearly state that when a fixed ebuild has been
> confirmed by two other people as working, we submit it to bugzilla. All
> the ebuild janitors do is make sure that the ebuilds use the USE flags
> to the maximum of their potential, and of course generally clean them up
> as there's a lot of stuff in some older ebuilds that is no longer valid.
> Your mail clearly showed that you had not read those guidelines..
> - Lovechild


--
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] Re: Ebuild Janitor Project

[foser]

On Sun, 2003-05-11 at 23:01, David Nielsen wrote:

> Okay, our guidelines clearly state that when a fixed ebuild has been
> confirmed by two other people as working, we submit it to bugzilla. All
> the ebuild janitors do is make sure that the ebuilds use the USE flags
> to the maximum of their potential, and of course generally clean them up
> as there's a lot of stuff in some older ebuilds that is no longer valid.
> Your mail clearly showed that you had not read those guidelines..

Why should we read every emerging sub projects documents? That's not
very realistic. Anyway, what defines an ebuild as working (this is a
problem in general) : does it build, does it build on a clean install
with any possible combination of USE flags, does it run fine for 1 month
on someones production server ?

Older ebuilds usually get phased out with time and cleaning is done on
new releases. Sure there will slip stuff trough now and then, nothing
major.

I had a quick peek at your ebuilds btw and i can tell you your
developers clearly do not read ChangeLog's for one, cause some things
that you do are certainly gonna give trouble.

What i'm afraid of here is that you and your project will give more
maintenance type of bugs which correct insignificant stuff, pressuring
an already pressured group of devs. And you guys happily define new USE
flags allover, which also isn't a good thing if you want to see your
changes integrated.

> The plan is NOT to fork portage, the plan is NOT to stage a coup, the
> plan is NOT to start a pissing contest with the developers... the plan
> is quite simple, to make portage even better.

A few ebuild corrections and a fork are whole different things. You have
obviously no idea what it takes to handle a full tree of ebuilds,
otherwise you wouldn't even have mentioned forking here in the first
place.

> Should we be so unlucky as to hit a major problem that we can't handle
> within the project, we will of course take contact to the person in
> charge, or mail gentoo-dev for input depending on the situation.

I'm not sure -dev is the right place for your problems with correcting
ebuild XYZ, it is already too high traffic for the general dev to keep
up with. Bugzilla or direct contact is probably best.

> We have already hit some cases that could be viewed as imperfections in
> the way flags are handled in Portage, and we have yet to receive any
> technical feedback on these issues.

Portage isn't perfect and there's a large TODO for it and i think we are
well aware of that. Developers have limited time, please take that in
mind when you want clarification about the feature you really need right
now/bug you encountered.

- foser


--
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] Re: Ebuild Janitor Project -- Any formalities tojoining it?

[foser]

On Sun, 2003-05-11 at 22:05, Marius Mauch wrote:
> What is the recommended way then to get in touch with the devs? We
> thought that sending a mail to -dev would be the best way. See
> Lovechilds mail as a RFC. Your mail had some good points, but I think it
> sounds quite harsh.

Bugzilla is the recommended way and i think that's mentioned in docs
(altough i haven't read one in ages), maybe a personal mail will do as
well. Devs got a lot of stuff to do usually and its hard to keep track
of that without something like bugzilla.

About being harsh : we're all adults here (well maybe not me, but hey),
we don't have to wrap up everything in soothing phrases do we ? It's not
politics or something. Just get to your point and don't quote 2 layers
of replies unneeded (pun intended).


> Ok, we all calm down a bit. I can't talk for the others, but I see the
> project as a testbed for ebuild modifications that can be tested by
> users before they get submitted to bugs.g.o. 

As long as you state _very clearly_ those ebuilds are unsupported by
Gentoo and it's development team. Even then it's possible we run into
problems caused by your ebuilds. Imagine how hard those are to track
down.

> As Lovechild has said there
> are also discussions about new use flags and other modifications to the
> current portage system. I missed the discussion on IRC but I saw that
> you were backed by other devs. Please notice that this projects has been
> started just for a few days so of course we haven't clarified all
> issues.

Well Gentoo has been around for a few years now and still hasn't reached
Distro Nirwana , it's a long way soulsearching for the right way to your
true perfect self. Don't think these issues you see, these ideas you
have, have never been brought up before. Just because you never saw them
doesn't mean they haven't been discussed and thought about.

That doesn't mean i don't want to support things you might bring up, but
you might find we have good reasons we do the things how we do them (you
don't say huh).

People sometimes come up to me and ask me how they can help. I always
tell them to go to our bugzilla and pick up a few bugs, look at them and
find a solution. Somehow it never happens, somehow i never see someone
fixing a bug and posting the solution to bugzilla. You might want to
rethink your approach and try help work down our amount of bugs first
instead of setting out with the intent to introduce more.

- foser


--
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] Re: Ebuild Janitor Project

[Spider]

[-- Attachment #1: Type: text/plain, Size: 1333 bytes --]

begin  quote
On 11 May 2003 23:33:20 +0200
foser <foser@foser.dyn.warande.net> wrote:

>  Anyway, what defines an ebuild as working (this is a
> problem in general) : does it build, does it build on a clean install
> with any possible combination of USE flags, does it run fine for 1
> month on someones production server ?
> 

Heads up those of you who go playing with stable (ie KEYWORDS="x86" )
packages.  Always make sure to run through all packages that -depend- on
it as well.. . Its a headache, but you'll need to check that the newly
introduced USE flag you just slapped on because that gnome, guile,
esound, jpeg, png,  support is really just optional... Yes?

It might not be so.  if a package Foo, depends on bar being built with
suopport for baz, making baz an optional thing is a no-no.  And even if
portage gets a way to track optional dependencies (foo depends on bar
with baz )   you'd still have to find all those builds. No, thats not
what testing is for, testing is not your bed to find the dependencies
you broke.  Thats what you are supposed to do.


Just a minor rant from somone who has seen a lot of people request some
USE flags without researching.


//Spider

-- 
begin  .signature
This is a .signature virus! Please copy me into your .signature!
See Microsoft KB Article Q265230 for more information.
end

[-- Attachment #2: Type: application/pgp-signature, Size: 189 bytes --]