[gentoo-dev] Please follow keywording policy

[gentoo-dev] Please follow keywording policy

[Jason Wever]

[-- Attachment #1: Type: text/plain, Size: 714 bytes --]

Hi All,

I'd like to ask that you all take the time to review the keywording
policy in the Developer Handbook.  In particular, I'd like to draw
your attention to the section on "Upgrading Ebuilds" [1].

People have been getting good lately at either dropping keywords for
no reason and/or failing to notify or file a bug with the arches
dropped as to why.  Chances are if you have done this for SPARC, you've
probably heard from me already.  If not, GenBot is on his way to your
location as we speak ;)

Now back to your regularly scheduled hack-fu...

[1] -
http://www.gentoo.org/proj/en/devrel/handbook/handbook.xml?part=2&chap=5#doc_chap5

Thanks,
-- 
Jason Wever
Gentoo/Sparc Team Co-Lead

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

Re: [gentoo-dev] Please follow keywording policy

[Hasan Khalil]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1


On Mar 8, 2005, at 19:55, Jason Wever wrote:

> Hi All,
>
> I'd like to ask that you all take the time to review the keywording
> policy in the Developer Handbook.  In particular, I'd like to draw
> your attention to the section on "Upgrading Ebuilds" [1].
>
> People have been getting good lately at either dropping keywords for
> no reason and/or failing to notify or file a bug with the arches
> dropped as to why.  Chances are if you have done this for SPARC, you've
> probably heard from me already.  If not, GenBot is on his way to your
> location as we speak ;)
>
> Now back to your regularly scheduled hack-fu...
>
> [1] -
> http://www.gentoo.org/proj/en/devrel/handbook/handbook.xml? 
> part=2&chap=5#doc_chap5
>
> Thanks,
> -- 
> Jason Wever
> Gentoo/Sparc Team Co-Lead
>

Amen to that. People been doing a great job dropping ppc-macos. Pretty  
please don't drop keywords? With strawberries on top?

- --
Hasan Khalil
Ebuild/Porting Co-Lead
Gentoo for Mac OS X

PGP Key: 0x707B8F18 on pgp.mit.edu
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.0 (Darwin)

iD8DBQFCLljEzsotBnB7jxgRAgdgAJ0esSxrC+HDUiQUi8D6gnJ3JZKVUwCeNT0c
Faos8+U8Dy9GqnV+WI10XMI=
=LtiH
-----END PGP SIGNATURE-----

--
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] Please follow keywording policy

[Alin Nastac]

[-- Attachment #1: Type: text/plain, Size: 977 bytes --]

I didn't dropped any keywords yet but I've been pretty close to that. 
See http://bugs.gentoo.org/show_bug.cgi?id=81702 for more info.

I think the reason people drop arches is laziness of some arch herds. 
C'mon people, how hard can it be to see if it builds right on your arch?

Jason Wever wrote:

>Hi All,
>
>I'd like to ask that you all take the time to review the keywording
>policy in the Developer Handbook.  In particular, I'd like to draw
>your attention to the section on "Upgrading Ebuilds" [1].
>
>People have been getting good lately at either dropping keywords for
>no reason and/or failing to notify or file a bug with the arches
>dropped as to why.  Chances are if you have done this for SPARC, you've
>probably heard from me already.  If not, GenBot is on his way to your
>location as we speak ;)
>
>Now back to your regularly scheduled hack-fu...
>
>[1] -
>http://www.gentoo.org/proj/en/devrel/handbook/handbook.xml?part=2&chap=5#doc_chap5
>
>Thanks,
>  
>


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 256 bytes --]

Re: [gentoo-dev] Please follow keywording policy

[Stephen P. Becker]

Alin Nastac wrote:
> I didn't dropped any keywords yet but I've been pretty close to that. 
> See http://bugs.gentoo.org/show_bug.cgi?id=81702 for more info.

Congratulations, you just took something that pissed you off personally 
and threw it up on the dev mailing list for the purpose of attempting to 
make a point where there is none.  Never mind that for well over a week, 
you joined #gentoo-mips at least 6 hours before any of us in there would 
even be awake, said only "mips team ping", and then logged off 5 minutes 
later.  You should have at least talked directly to somebody before 
getting so pissed off.  Remember, when *you* want something, *we're* not 
likely to come track you down.

> I think the reason people drop arches is laziness of some arch herds.

Laziness?  I can't speak for other herds, but when stuff like this gets 
thrown on the backburner with respect to mips, it is because most of us 
don't even come close to working on gentoo full time.  And then, each of 
us has a specific area that we take care of.  I, for example, deal 
pretty much only with X stuff.  Most anything else I consider out of 
bounds since I either don't know how to test a certain package, or I 
have no way of testing it even if I wanted to, which leads me to the 
next point...

> C'mon people, how hard can it be to see if it builds right on your arch?

So you are suggesting that we should mark something stable if it simply 
compiles?  I really hope you don't apply this lazy method of QA to 
everything you maintain, else I fear for the users installing those 
packages.  This is the main reason we didn't touch that package, because 
none of us had the ability to test pppoe.  If you had actually made an 
attempt to talk to one of us when we were awake before running your 
mouth, you would have known this.  In fact, we probably would have just 
given you permission to remove the keyword assuming that no repoman 
breakage resulted.

Steve


--
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] Please follow keywording policy

[Alin Nastac]

[-- Attachment #1: Type: text/plain, Size: 2568 bytes --]

Stephen P. Becker wrote:

> Alin Nastac wrote:
>
>> I didn't dropped any keywords yet but I've been pretty close to that. 
>> See http://bugs.gentoo.org/show_bug.cgi?id=81702 for more info.
>
>
> Congratulations, you just took something that pissed you off 
> personally and threw it up on the dev mailing list for the purpose of 
> attempting to make a point where there is none.  Never mind that for 
> well over a week, you joined #gentoo-mips at least 6 hours before any 
> of us in there would even be awake, said only "mips team ping", and 
> then logged off 5 minutes later.  You should have at least talked 
> directly to somebody before getting so pissed off.  Remember, when 
> *you* want something, *we're* not likely to come track you down.
>
Ok, I could understand that none of the mips team are online when I 
want, but how about the bug mentioned above? One month I 
waited/begged/threat!!! For what? For some lousy script updates, which 
is not arch dependent anyway!


>> C'mon people, how hard can it be to see if it builds right on your arch?
>
>
> So you are suggesting that we should mark something stable if it 
> simply compiles?  I really hope you don't apply this lazy method of QA 
> to everything you maintain, else I fear for the users installing those 
> packages.  This is the main reason we didn't touch that package, 
> because none of us had the ability to test pppoe.  If you had actually 
> made an attempt to talk to one of us when we were awake before running 
> your mouth, you would have known this.  In fact, we probably would 
> have just given you permission to remove the keyword assuming that no 
> repoman breakage resulted.
>
Not every package that I maintain is tested by me! How can I test a dsl 
driver when I have no such device!
I apply the good old rule that if a package has no open bugs for a 
month, it will be marked as stable.

If I would do things you way, in this moment net-dialup would have at 
least 100 opened bugs with no one carrying about them. I urge you to 
look in bugzilla to see how many rightfully complaints are there 
regarding my work. Not as if I consider less about a dev who made a 
mistake (only who do nothing, does no mistakes)...

It is OK to want to make more than a simple compile test, but from this 
to doing nothing when a fellow dev ask you to IS a big distance! A 
convenient excuse, nothing more...

I didn't wanted to get so involved in this bug, but invalid bug reports 
started by the transition from -r2 are killing me. In rest, what do I 
care that your arch is outdated!


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 256 bytes --]

Re: [gentoo-dev] Please follow keywording policy

[Ciaran McCreesh]

[-- Attachment #1: Type: text/plain, Size: 482 bytes --]

On Wed, 09 Mar 2005 16:50:22 +0200 Alin Nastac <mrness@gentoo.org>
wrote:
| Not every package that I maintain is tested by me! How can I test a
| dsl  driver when I have no such device!

Simple. You don't maintain dsl drivers.

Just because *you* have a shoddy QA attitude doesn't mean the rest of us
should.

-- 
Ciaran McCreesh : Gentoo Developer (Vim, Fluxbox, shell tools)
Mail            : ciaranm at gentoo.org
Web             : http://dev.gentoo.org/~ciaranm


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

Re: [gentoo-dev] Please follow keywording policy

[Stephen P. Becker]

> Ok, I could understand that none of the mips team are online when I 
> want, but how about the bug mentioned above? One month I 
> waited/begged/threat!!! For what? For some lousy script updates, which 
> is not arch dependent anyway!

There are lots of open bugs, sometimes they slip under the radar.  And 
*never* assume scripts are arch independant.  See more below...

> Not every package that I maintain is tested by me! How can I test a dsl 
> driver when I have no such device!

Do you have somebody in your herd test it that does have such a device? 
  If not, you shouldn't be maintaining it.

> I apply the good old rule that if a package has no open bugs for a 
> month, it will be marked as stable.

This if fine if you can vouch that it works.  None of us could say 
rp-pppoe works.  What is more, just because it works on x86 doesn't mean 
there aren't problems with other arches.  Even things like 
perl/python/<insert favorite scripting language here> scripts are 
included in this statement.  If I remember correctly, ciaran even ran 
into a non-portable vim script at one point.

That said, since you are the package maintainer, you should be 
practically the world's expert on this package.  You should know if 
there are potential endian problems.  You should be talking to upstream 
to find these sorts of things out.  If you have full confidence that it 
should be stable on our arch, mark it stable.  Seeing as you see fit to 
break policy and mark your dsl driver package stable even though you 
can't test it, then how is this any different?  Keep in mind that I'm 
pretty sure there are developer machines available for (almost) every 
arch now.  Package maintainers should use this resource for testing 
purposes.

> If I would do things you way, in this moment net-dialup would have at 
> least 100 opened bugs with no one carrying about them. I urge you to 
> look in bugzilla to see how many rightfully complaints are there 
> regarding my work. Not as if I consider less about a dev who made a 
> mistake (only who do nothing, does no mistakes)...

You are missing the point.

> It is OK to want to make more than a simple compile test, but from this 
> to doing nothing when a fellow dev ask you to IS a big distance! A 
> convenient excuse, nothing more...

Convenient excuse for what?  Do you think we singled you out and said, 
"hey, I don't like that mrness...we're just going to ignore his package" ?

> I didn't wanted to get so involved in this bug, but invalid bug reports 
> started by the transition from -r2 are killing me. In rest, what do I 
> care that your arch is outdated!

I don't quite see what the big deal is anyway.  When I looked, the last 
stable version of that package was from last november or so.  If we were 
a year out of date or something, that might be a different story.  We do 
have a script that emails all of the mips team with outdated packages 
from time to time, however we have a shortage of devs and time, so 
non-crucial stuff that none of the mips team uses like rp-pppoe 
typically gets pushed back.

Steve



--
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] Please follow keywording policy

[Chris Gianelloni]

[-- Attachment #1: Type: text/plain, Size: 2050 bytes --]

On Wed, 2005-03-09 at 09:38 +0200, Alin Nastac wrote:
> I think the reason people drop arches is laziness of some arch herds. 
> C'mon people, how hard can it be to see if it builds right on your arch?

What arch do you use?  Maybe we should trade in your box for a nice
sparc32 or mips box and see what you have to say then.  Remember that in
many of these arches, all the machines are still measured in Megahertz
and not Gigahertz, and they are quite old.

The arch teams are doing their jobs quite well, don't try to push blame
onto them.  They shouldn't go around marking something stable just
because it builds and should test it.  If they have no way of testing
it, then they don't need to stabilize it.  It won't kill you to have a
single older ebuild in the tree for an arch.  Either that, or you can
remove the keywords, as Jason mentioned, and file a bug against the
package to the mips team so they are aware that keywords have been
dropped from the package and that it will need testing to be
re-keyworded.  Looking over that bug, it really looks like you flipped
out over nothing.  It took them a week to respond.  That isn't very
long, at all.

> Jason Wever wrote:
> 
> >Hi All,
> >
> >I'd like to ask that you all take the time to review the keywording
> >policy in the Developer Handbook.  In particular, I'd like to draw
> >your attention to the section on "Upgrading Ebuilds" [1].
> >
> >People have been getting good lately at either dropping keywords for
> >no reason and/or failing to notify or file a bug with the arches
> >dropped as to why.  Chances are if you have done this for SPARC, you've
> >probably heard from me already.  If not, GenBot is on his way to your
> >location as we speak ;)
> >
> >Now back to your regularly scheduled hack-fu...
> >
> >[1] -
> >http://www.gentoo.org/proj/en/devrel/handbook/handbook.xml?part=2&chap=5#doc_chap5
> >
> >Thanks,
> >  
> >
> 
-- 
Chris Gianelloni
Release Engineering - Strategic Lead/QA Manager
Games - Developer
Gentoo Linux

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

Re: [gentoo-dev] Please follow keywording policy

[Alin Nastac]

[-- Attachment #1: Type: text/plain, Size: 670 bytes --]

Ciaran McCreesh wrote:

>On Wed, 09 Mar 2005 16:50:22 +0200 Alin Nastac <mrness@gentoo.org>
>wrote:
>| Not every package that I maintain is tested by me! How can I test a
>| dsl  driver when I have no such device!
>
>Simple. You don't maintain dsl drivers.
>
>  
>
And who else would like to step up as mantainer? I don't do that for fun 
(at my age - 10 years of experience - you stop being amused by this sort 
of things), I do it because someone has to and because I love this distro.

>Just because *you* have a shoddy QA attitude doesn't mean the rest of us
>should.
>  
>

Well, I will not ask you for help, that's for sure. Not after your 
categoric "no" on irc.

[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 256 bytes --]

Re: [gentoo-dev] Please follow keywording policy

[Alin Nastac]

[-- Attachment #1: Type: text/plain, Size: 1364 bytes --]

Chris Gianelloni wrote:

>On Wed, 2005-03-09 at 09:38 +0200, Alin Nastac wrote:
>  
>
>>I think the reason people drop arches is laziness of some arch herds. 
>>C'mon people, how hard can it be to see if it builds right on your arch?
>>    
>>
>
>What arch do you use?  Maybe we should trade in your box for a nice
>sparc32 or mips box and see what you have to say then.  Remember that in
>many of these arches, all the machines are still measured in Megahertz
>and not Gigahertz, and they are quite old.
>
>The arch teams are doing their jobs quite well, don't try to push blame
>onto them.  They shouldn't go around marking something stable just
>because it builds and should test it.  If they have no way of testing
>it, then they don't need to stabilize it.  It won't kill you to have a
>single older ebuild in the tree for an arch.  Either that, or you can
>remove the keywords, as Jason mentioned, and file a bug against the
>package to the mips team so they are aware that keywords have been
>dropped from the package and that it will need testing to be
>re-keyworded.  Looking over that bug, it really looks like you flipped
>out over nothing.  It took them a week to respond.  That isn't very
>long, at all.
>  
>
Hmm.. it looks like I've overreacted. I was convinced that my request 
have been ignored for one month.
My apologies to mips team and all...

[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 256 bytes --]

Re: [gentoo-dev] Please follow keywording policy

[Ciaran McCreesh]

[-- Attachment #1: Type: text/plain, Size: 1310 bytes --]

On Wed, 09 Mar 2005 19:38:10 +0200 Alin Nastac <mrness@gentoo.org>
wrote:
| Ciaran McCreesh wrote:
| >On Wed, 09 Mar 2005 16:50:22 +0200 Alin Nastac <mrness@gentoo.org>
| >wrote:
| >| Not every package that I maintain is tested by me! How can I test a
| >| dsl  driver when I have no such device!
| >
| >Simple. You don't maintain dsl drivers.
| >
| And who else would like to step up as mantainer? I don't do that for
| fun  (at my age - 10 years of experience - you stop being amused by
| this sort  of things), I do it because someone has to and because I
| love this distro.

Stick out a request on -dev. I bet someone here has a dsl device...

| >Just because *you* have a shoddy QA attitude doesn't mean the rest of
| >us should.
| 
| Well, I will not ask you for help, that's for sure. Not after your 
| categoric "no" on irc.

You were asking me to test an app which I don't have the hardware for.
What do you expect? "I can't test it but I'll mark it stable anyway"? I
hope no-one is still seriously taking the whole "maintainer arch" thing
seriously after having seen how some of the x86 developers behave...

-- 
Ciaran McCreesh : Gentoo Developer (Vim, Fluxbox, shell tools)
Mail            : ciaranm at gentoo.org
Web             : http://dev.gentoo.org/~ciaranm


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

Re: [gentoo-dev] Please follow keywording policy

[Alin Nastac]

[-- Attachment #1: Type: text/plain, Size: 482 bytes --]

Ciaran McCreesh wrote:

>Stick out a request on -dev. I bet someone here has a dsl device...
>  
>
Okay, I buy it.

a) Devs, who has a fcdsl, fcdsl2, fcdslsl, fcdslusb and/or fcdslslusb 
(AVM FRITZ!Card DSL)? Step up and take net-dialup/fcdsl under your wing!
b) Who has a device that works with net-dialup/hcfusbmodem and is 
willing to become its maintainer?
c)...

In fact, who has any device that would work with some driver from 
net-dialup? I don't have _any_ device at all!


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 256 bytes --]

Re: [gentoo-dev] Please follow keywording policy

[Stefan Schweizer]

On Wed, 09 Mar 2005 20:56:51 +0200, Alin Nastac <mrness@gentoo.org> wrote:
> Okay, I buy it.
> 
> a) Devs, who has a fcdsl, fcdsl2, fcdslsl, fcdslusb and/or fcdslslusb
> (AVM FRITZ!Card DSL)? Step up and take net-dialup/fcdsl under your wing!
I got two of these donated by AVM (fcdslusb and fcdslsl). And
additionally a fcusb for net-dialup/fritzcapi
I think all the AVM-drivers in the tree are covered by me.

I dont think that it means that you are a bad maintainer when you do
not own the hardware. But if you dont like the work you should not do
it, please tell us then, I am ready to take over as much as I can to
help you.
> b) Who has a device that works with net-dialup/hcfusbmodem and is
> willing to become its maintainer?
sorry, I dont own a device .. nor does any developer I think, should
these ebuilds be not maintained then?
Imo when something cannot be tested it should be at least on its
latest version and in the tree and bugs that come up should be fixed.
And that can a maintainer do without owning the device.
> c)...
> 
> In fact, who has any device that would work with some driver from
> net-dialup? I don't have _any_ device at all!
Do you have a dsl-connection?
How do you connect to the internet?

Regards,
Stefan
--
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] Please follow keywording policy

[Stefan Schweizer]

On Wed, 09 Mar 2005 20:56:51 +0200, Alin Nastac <mrness@gentoo.org> wrote:
> Okay, I buy it.
> 
> a) Devs, who has a fcdsl, fcdsl2, fcdslsl, fcdslusb and/or fcdslslusb
> (AVM FRITZ!Card DSL)? Step up and take net-dialup/fcdsl under your wing!
I got two of these donated by AVM (fcdslusb and fcdslsl). And
additionally a fcusb for net-dialup/fritzcapi
I think all the AVM-drivers in the tree are covered by me.

I dont think that it means that you are a bad maintainer when you do
not own the hardware. But if you dont like the work you should not do
it, please tell us then, I am ready to take over as much as I can to
help you.
> b) Who has a device that works with net-dialup/hcfusbmodem and is
> willing to become its maintainer?
sorry, I dont own a device .. nor does any developer I think, should
these ebuilds be not maintained then?
Imo when something cannot be tested it should be at least on its
latest version and in the tree and bugs that come up should be fixed.
And that can a maintainer do without owning the device.
> c)...
> 
> In fact, who has any device that would work with some driver from
> net-dialup? I don't have _any_ device at all!
Do you have a dsl-connection?
How do you connect to the internet?

Regards,
Stefan
--
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] Please follow keywording policy

[Patrick Lauer]

[-- Attachment #1: Type: text/plain, Size: 811 bytes --]

On Wed, 2005-03-09 at 20:56 +0200, Alin Nastac wrote:
> a) Devs, who has a fcdsl, fcdsl2, fcdslsl, fcdslusb and/or fcdslslusb 
> (AVM FRITZ!Card DSL)?
I had a fcdslsl for about ... well... two days (and no Internet in that
time :-) ).
It's about as crappy as most "softmodems". I'm back to an external
pppoe-based DSL modem since it just works.

I can only imagine the horrors of getting the others to work :-)

>  Step up and take net-dialup/fcdsl under your wing!
No thanks. AVM has gone from "very good" to "cheap broken stuff" in a
very short time :-(

> In fact, who has any device that would work with some driver from 
> net-dialup? I don't have _any_ device at all!
I still have a fcdslsl and an external pppoe-based DSL modem lying 
around for any interested devs.

wkr,
Patrick


[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 198 bytes --]

Re: [gentoo-dev] Please follow keywording policy

[Chris Gianelloni]

[-- Attachment #1: Type: text/plain, Size: 1079 bytes --]

On Wed, 2005-03-09 at 18:38 +0000, Ciaran McCreesh wrote:
> You were asking me to test an app which I don't have the hardware for.
> What do you expect? "I can't test it but I'll mark it stable anyway"? I
> hope no-one is still seriously taking the whole "maintainer arch" thing
> seriously after having seen how some of the x86 developers behave...

They aren't "x86 developers" just because they're on an x86 anymore than
I'm an amd64 developer or a sparc developer because I'm on an amd64 or a
sparc.  After all, I do some of my "development" from my PPC box,
running MacOS X, but I'm neither a PPC, nor a MacOS developer.  Just
because someone is a developer and happens to use x86, don't try to make
this an "x86 versus $arch" problem, when it is really a "package
maintainer versus $arch" one.

That all being said, it would be nice if more people used arches other
than x86 every once in a while just so they could see how different
things can be.

-- 
Chris Gianelloni
Release Engineering - Strategic Lead/QA Manager
Games - Developer
Gentoo Linux

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

Re: [gentoo-dev] Please follow keywording policy

[Stefan Schweizer]

On Wed, 09 Mar 2005 20:18:35 +0100, Patrick Lauer <patrick@gentoo.org> wrote:
> I had a fcdslsl for about ... well... two days (and no Internet in that
> time :-) ).
The fcdslsl driver in portage was not that advanced by then I guess.
> It's about as crappy as most "softmodems". I'm back to an external
> pppoe-based DSL modem since it just works.
Quality of softmodems depends on the drivers .. if you use crappy
drivers you get disconnects every half an hour ..

The disadvantage of my external modem is that it has a router and some
other things included which I cant use(only one ethernet
port+passwordprotected by provider), its another box in the crowded
computer-room and it nevertheless needs an ethernet-card in the
computer. I am almost sure power consumption does also matter for that
extra 24/7 box.

> I can only imagine the horrors of getting the others to work :-)

no horrors here .. was actually pretty easy 
ebuild /path/to config
sets everything up.

> No thanks. AVM has gone from "very good" to "cheap broken stuff" in a
> very short time :-(
So everything form AVM has gone to "cheap broken stuff", because you
were not able to set up your dsl-card? Please give us some moer
examples .. I think they are actually pretty cool, especially the new
fritzboxfons.
> I still have a fcdslsl and an external pppoe-based DSL modem lying
> around for any interested devs.

What pppoe-based DSL modem? Is it an usb-one that needs special
drivers? Or just one of the ethernet-connected that works
out-of-the-box?

Regards,
Stefan
--
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] Please follow keywording policy

[Patrick Lauer]

[-- Attachment #1: Type: text/plain, Size: 2024 bytes --]

On Wed, 2005-03-09 at 20:43 +0100, Stefan Schweizer wrote:
> On Wed, 09 Mar 2005 20:18:35 +0100, Patrick Lauer <patrick@gentoo.org> wrote:
> > I had a fcdslsl for about ... well... two days (and no Internet in that
> > time :-) ).
> The fcdslsl driver in portage was not that advanced by then I guess.
The official AVM driver was not yet working on linux at that time ...

> > It's about as crappy as most "softmodems". I'm back to an external
> > pppoe-based DSL modem since it just works.
> Quality of softmodems depends on the drivers .. if you use crappy
> drivers you get disconnects every half an hour ..
Hehe. If you live in Germany, you get disconnects every [5 ... 1440] minutes anyway :-)


> > No thanks. AVM has gone from "very good" to "cheap broken stuff" in a
> > very short time :-(
> So everything form AVM has gone to "cheap broken stuff", because you
> were not able to set up your dsl-card?
Well ... their latest offerings I could test were extremely
windows-centric and did not work to my satisfaction. And th fcdslsl is a
cheap DSP with some PCI interface glue ...

>  Please give us some moer
> examples .. I think they are actually pretty cool, especially the new
> fritzboxfons.
Haven't tested those yet. I hope they are better!

> > I still have a fcdslsl and an external pppoe-based DSL modem lying
> > around for any interested devs.
> 
> What pppoe-based DSL modem? Is it an usb-one that needs special
> drivers? Or just one of the ethernet-connected that works
> out-of-the-box?
It's an external ethernet-based thingy, belgian produce (so it might not
be fully standards compliant) for an analog+DSL connection (not ISDN)

I tend to keep away from those DSL-over-USB-over-SATA-over-WLAN things.
It's unneeded complexity that tends to make these things really hard to
use.

Note that I'm not saying that AVM produces cr*p. I'm only saying that
their old ISDN products were really good and their new stuff has been
rather frustrating for me.

Patrick

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 198 bytes --]

Re: [gentoo-dev] Please follow keywording policy

[Luis F. Araujo]

Stefan Schweizer wrote:

>On Wed, 09 Mar 2005 20:56:51 +0200, Alin Nastac <mrness@gentoo.org> wrote:
>  
>
>>Okay, I buy it.
>>
>>a) Devs, who has a fcdsl, fcdsl2, fcdslsl, fcdslusb and/or fcdslslusb
>>(AVM FRITZ!Card DSL)? Step up and take net-dialup/fcdsl under your wing!
>>    
>>
>I got two of these donated by AVM (fcdslusb and fcdslsl). And
>additionally a fcusb for net-dialup/fritzcapi
>I think all the AVM-drivers in the tree are covered by me.
>
>I dont think that it means that you are a bad maintainer when you do
>not own the hardware. But if you dont like the work you should not do
>it, please tell us then, I am ready to take over as much as I can to
>help you.
>  
>
>>b) Who has a device that works with net-dialup/hcfusbmodem and is
>>willing to become its maintainer?
>>    
>>
>sorry, I dont own a device .. nor does any developer I think, should
>these ebuilds be not maintained then?
>Imo when something cannot be tested it should be at least on its
>latest version and in the tree and bugs that come up should be fixed.
>And that can a maintainer do without owning the device.
>  
>
>
Just a brief comment, we should keep ~arch for those drivers/applications
that still can't be tested for X or Y reason, and for informative purpose,
we also could include this information in metada.xml for example. (for users
to be aware of the current situation mainly, and at the same time for a 
_way_
to call people interested on mantaining the package).

Cheers,


--
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] Please follow keywording policy

[Alin Nastac]

[-- Attachment #1: Type: text/plain, Size: 2725 bytes --]

Stefan Schweizer wrote:

>On Wed, 09 Mar 2005 20:56:51 +0200, Alin Nastac <mrness@gentoo.org> wrote:
>  
>
>>Okay, I buy it.
>>
>>a) Devs, who has a fcdsl, fcdsl2, fcdslsl, fcdslusb and/or fcdslslusb
>>(AVM FRITZ!Card DSL)? Step up and take net-dialup/fcdsl under your wing!
>>    
>>
>I got two of these donated by AVM (fcdslusb and fcdslsl). And
>additionally a fcusb for net-dialup/fritzcapi
>I think all the AVM-drivers in the tree are covered by me.
>
>I dont think that it means that you are a bad maintainer when you do
>not own the hardware. But if you dont like the work you should not do
>it, please tell us then, I am ready to take over as much as I can to
>help you.
>  
>
You already help me a lot, Stefan, and I thank you for that.

I don't really intend to drop my jobs to others - this isn't me, I take 
my responsibilities very seriously. But if you think you'll be doing a 
better job by taking it, please, be my guest. I don't want to appear as 
the guy who broked Gentoo's QA procedures...

I'm only annoyed by the bad attitude of some devs who will get involved 
only what suits them, forgetting that if they would not help, no one 
will. Btw, what is the sense of ~arch if not "testing"? No gentooer 
expects from a ~arch ebuild to be stable, so the sky would not fall if 
you made a mistake and release it under this keyword. When I hear "I 
cannot mark foo library as ~arch because I don't know how to test it" 
smells like excuse to me.

As for QA... does anyone think we *can* have proper QA procedures, with 
our release speed and decentralized development model? And with only ... 
350 devs from which God knows how many are still active?  :-D
Who thinks that clearly doesn't have a clue what QA means. It is 
practically impossible to test every combination of ebuilds/USE/CFLAGS 
so all we do is a surface test, letting the burden of testing on the 
shoulders of our users.
Despite of our unorthodox development process, many people believes 
(including me) that our distro surclass traditional ones and is 
generally more stable (go figure!).

Maybe I'm too exigent, but I only ask from people to do what I do : be 
genuinely interested in helping the devs who need it. Heck, I always try 
to help any gentooer, dev or not. We all have our little systems because 
our predecesors have worked on it, not because they  sit down and 
debated whether to mark foo ebuild as ~arch or not.

>>c)...
>>
>>In fact, who has any device that would work with some driver from
>>net-dialup? I don't have _any_ device at all!
>>    
>>
>Do you have a dsl-connection?
>How do you connect to the internet?
>  
>
I do have a dsl connection, but it is through a modem with an eth 
interface, so I need no driver.

[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 256 bytes --]

Re: [gentoo-dev] Please follow keywording policy

[Stephen P. Becker]

> I don't really intend to drop my jobs to others - this isn't me, I take 
> my responsibilities very seriously. But if you think you'll be doing a 
> better job by taking it, please, be my guest. I don't want to appear as 
> the guy who broked Gentoo's QA procedures...

Nobody is asking you to drop your job.  However, what people are saying 
is if you don't have hardware, then how can you properly maintain a 
driver for it?  If I didn't have any mips machines, how could I be a 
member of the mips team?  Instead of trying to maintain a bunch of 
drivers that you can't test, why don't you recruit some more devs for 
the dialup herd who do have such hardware?  You can create a team that 
coordinates through you to ensure good QA for these drivers.

> 
> I'm only annoyed by the bad attitude of some devs who will get involved 
> only what suits them, forgetting that if they would not help, no one 
> will.

Again, other developers can only get involved (in this situation) if 
they have hardware.

> Btw, what is the sense of ~arch if not "testing"? No gentooer 
> expects from a ~arch ebuild to be stable, so the sky would not fall if 
> you made a mistake and release it under this keyword. When I hear "I 
> cannot mark foo library as ~arch because I don't know how to test it" 
> smells like excuse to me.

Either you are confused or you are trying to turn this into a subtle 
troll.  This fork of the discussion started in reference to *stable* 
keywording with respect to the rp-pppoe bug you used as an example.  I 
usually have no problem marking something ~arch if it compiles, since 
~arch just means it is a candidate for possibly becoming stable some 
time in the future.  Anyone using ~arch keywords should be prepared for 
a bumpy ride.  However, once that ebuild goes stable, in theory it 
should JustWork(TM) with no problems.  In this particular case, none of 
the mips team could vouch that rp-pppoe JustWorks(TM).

> As for QA... does anyone think we *can* have proper QA procedures, with 
> our release speed and decentralized development model? 

Sure.

> And with only ... 
> 350 devs from which God knows how many are still active?  :-D
> Who thinks that clearly doesn't have a clue what QA means. It is 
> practically impossible to test every combination of ebuilds/USE/CFLAGS 
> so all we do is a surface test, letting the burden of testing on the 
> shoulders of our users.

This is what ~arch is for.

> Despite of our unorthodox development process, many people believes 
> (including me) that our distro surclass traditional ones and is 
> generally more stable (go figure!).

I'm not sure what you are trying to say here.  Stating that we have an 
excellent distro doesn't mean that we can bypass QA policy.

> Maybe I'm too exigent, but I only ask from people to do what I do : be 
> genuinely interested in helping the devs who need it. Heck, I always try 
> to help any gentooer, dev or not. We all have our little systems because 
> our predecesors have worked on it, not because they  sit down and 
> debated whether to mark foo ebuild as ~arch or not.

Again, you are referring to ~arch, which is not what the original 
problem was with.


Steve
--
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] Please follow keywording policy

[Jason Wever]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On Thu, 10 Mar 2005, Alin Nastac wrote:

> I'm only annoyed by the bad attitude of some devs who will get involved only 
> what suits them, forgetting that if they would not help, no one will. Btw, 
> what is the sense of ~arch if not "testing"? No gentooer expects from a ~arch 
> ebuild to be stable, so the sky would not fall if you made a mistake and 
> release it under this keyword. When I hear "I cannot mark foo library as 
> ~arch because I don't know how to test it" smells like excuse to me.

When a package is marked ~arch, it's also supposed to work (or have worked 
in a previous version).

Also it may help to look at things from an arch maintainers points of 
view.  We are responsible for essentially maintaining almost every package 
in the tree for our architecture.  When we ask for testing criteria with 
regards to a request for keywording, it's because it makes both of our 
lives easier and helps to improve QA as we arch maintainers don't know a 
lot about your application and don't have the time to get to know it like 
you do.  This is especially helpful when dealing with libraries.

Cheers,
- -- 
Jason Wever
Gentoo/Sparc Co-Team Lead
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.6 (GNU/Linux)

iD8DBQFCL4iCdKvgdVioq28RAhqxAKC2MDFXkkHaX97d2FyX22H4V7Q0ZACgg9Fm
SCaOh/OJ0fjnnDZVZRutG9A=
=snEB
-----END PGP SIGNATURE-----
--
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] Please follow keywording policy

[Aron Griffis]

[-- Attachment #1: Type: text/plain, Size: 1342 bytes --]

Alin Nastac wrote:	[Wed Mar 09 2005, 05:57:15PM EST]
> No gentooer 
> expects from a ~arch ebuild to be stable, so the sky would not fall if 
> you made a mistake and release it under this keyword. When I hear "I 
> cannot mark foo library as ~arch because I don't know how to test it" 
> smells like excuse to me.

Earlier you said that you mark ebuilds stable after 30 days with no
bugs.  Now you're suggesting that marking ~arch by mistake isn't a big
deal.  I don't wish to beat up on you further, but this viewpoint can
lead to stable ebuilds in the tree that don't even build over the
course of one month (other than February ;-)

> As for QA... does anyone think we *can* have proper QA procedures,
> with our release speed and decentralized development model? And with
> only ...  350 devs from which God knows how many are still active?
> :-D Who thinks that clearly doesn't have a clue what QA means. It is
> practically impossible to test every combination of
> ebuilds/USE/CFLAGS so all we do is a surface test, letting the
> burden of testing on the shoulders of our users.

You're right, we don't have real QA yet.  But it's still better to try
the best we can rather than marking things blindly on the basis that
"we don't have QA anyway"

Regards,
Aron

--
Aron Griffis
Gentoo Linux Developer


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

Re: [gentoo-dev] Please follow keywording policy

[Daniel Goller]

[-- Attachment #1: Type: text/plain, Size: 894 bytes --]

Donnie Berkholz wrote:
> Stephen P. Becker wrote:
> | Nobody is asking you to drop your job.  However, what people are saying
> | is if you don't have hardware, then how can you properly maintain a
> | driver for it?  If I didn't have any mips machines, how could I be a
> | member of the mips team?  Instead of trying to maintain a bunch of
> | drivers that you can't test, why don't you recruit some more devs for
> | the dialup herd who do have such hardware?  You can create a team that
> | coordinates through you to ensure good QA for these drivers.
>
> There is a fairly common case that a package exists that Gentoo _should_
> have in the tree, but those people with the hardware are unable to
> become devs because of time commitments or other reasons. I see no
> reason a dev can't collaborate with a non-dev on testing packages.
--
gentoo-dev@gentoo.org mailing list

Amen to that.


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 256 bytes --]

Re: [gentoo-dev] Please follow keywording policy

[Donnie Berkholz]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Stephen P. Becker wrote:
| Nobody is asking you to drop your job.  However, what people are saying
| is if you don't have hardware, then how can you properly maintain a
| driver for it?  If I didn't have any mips machines, how could I be a
| member of the mips team?  Instead of trying to maintain a bunch of
| drivers that you can't test, why don't you recruit some more devs for
| the dialup herd who do have such hardware?  You can create a team that
| coordinates through you to ensure good QA for these drivers.

There is a fairly common case that a package exists that Gentoo _should_
have in the tree, but those people with the hardware are unable to
become devs because of time commitments or other reasons. I see no
reason a dev can't collaborate with a non-dev on testing packages.
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.0 (GNU/Linux)
Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org

iD8DBQFCL5m8XVaO67S1rtsRAsBnAJ4hjuxGpxQFbucqEeDFPrcXwIFmigCgu9rF
dUwrdx5deoyaBkWVi/fNv+Q=
=1+qT
-----END PGP SIGNATURE-----
--
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] Please follow keywording policy

[Jon Portnoy]

On Wed, Mar 09, 2005 at 08:56:51PM +0200, Alin Nastac wrote:
> Ciaran McCreesh wrote:
> 
> >Stick out a request on -dev. I bet someone here has a dsl device...
> > 
> >
> Okay, I buy it.
> 
> a) Devs, who has a fcdsl, fcdsl2, fcdslsl, fcdslusb and/or fcdslslusb 
> (AVM FRITZ!Card DSL)? Step up and take net-dialup/fcdsl under your wing!
> b) Who has a device that works with net-dialup/hcfusbmodem and is 
> willing to become its maintainer?
> c)...
> 
> In fact, who has any device that would work with some driver from 
> net-dialup? I don't have _any_ device at all!
> 

Then how can we know anything in net-dialup actually works? Sounds like 
net-dialup is sorely lacking devs with the relevant hardware.

-- 
Jon Portnoy
avenj/irc.freenode.net
--
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] Please follow keywording policy

[Grant Goodyear]

[-- Attachment #1: Type: text/plain, Size: 1236 bytes --]

Alin Nastac wrote: [Wed Mar 09 2005, 04:57:15PM CST]
> Btw, what is the sense of ~arch if not "testing"? No gentooer 
> expects from a ~arch ebuild to be stable, so the sky would not fall if 
> you made a mistake and release it under this keyword. When I hear "I 
> cannot mark foo library as ~arch because I don't know how to test it" 
> smells like excuse to me.

*Sigh*  The meaning of ~arch is that, at a minimum, the package works
for the person who keyworded it (or, in some cases, worked for a trusted
user on whose behalf the package was keyworded).   In other words, the
dev believes that the package works, and that belief is based on
evidence, not just wishful thinking.  An "arch" keyword means that there
is considerable evidence that the package works for multiple people.
Packages that might work, but also might not, ideally should not be in
the tree at all, but could reasonably be package.mask'ed if testing is
imminent.

I'm pretty sure that I'm spouting the company line, here, but feel free
to correct me if I'm wrong.

-g2boojum-
-- 
Grant Goodyear	
Gentoo Developer
g2boojum@gentoo.org
http://www.gentoo.org/~g2boojum
GPG Fingerprint: D706 9802 1663 DEF5 81B0  9573 A6DC 7152 E0F6 5B76

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

Re: [gentoo-dev] Please follow keywording policy

[Alin Nastac]

[-- Attachment #1: Type: text/plain, Size: 1450 bytes --]

Aron Griffis wrote:

>Alin Nastac wrote:	[Wed Mar 09 2005, 05:57:15PM EST]
>  
>
>>No gentooer 
>>expects from a ~arch ebuild to be stable, so the sky would not fall if 
>>you made a mistake and release it under this keyword. When I hear "I 
>>cannot mark foo library as ~arch because I don't know how to test it" 
>>smells like excuse to me.
>>    
>>
>
>Earlier you said that you mark ebuilds stable after 30 days with no
>bugs.  Now you're suggesting that marking ~arch by mistake isn't a big
>deal.  I don't wish to beat up on you further, but this viewpoint can
>lead to stable ebuilds in the tree that don't even build over the
>course of one month (other than February ;-)
>
>  
>
I always test unpack/compile/install functionality of the ebuild before 
submittion, even if the submittion means only a script change.
So, from my point of view, every ebuild of mine at least must get 
installed. Of course, it doesn't mean that install process will succeed 
on any other machines than mine...
I'm only saying that no one will take your head if you do a mistake, but 
when you're not doing nothing because you are fearing of a mistake - 
this is problem. To err is human, you know.
As for marking an ebuild stable over a month, this is the ideal case. On 
average, I mark an ebuild stable on x86 after 2 months.

Again, I want to emphasize that I didn't break anything in portage tree, 
even if I have a "shoddy" attitude regarding Gentoo's QA.

[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 256 bytes --]

Re: [gentoo-dev] Please follow keywording policy

[Alin Nastac]

[-- Attachment #1: Type: text/plain, Size: 1433 bytes --]

Jon Portnoy wrote:

>On Wed, Mar 09, 2005 at 08:56:51PM +0200, Alin Nastac wrote:
>  
>
>>Ciaran McCreesh wrote:
>>
>>    
>>
>>>Stick out a request on -dev. I bet someone here has a dsl device...
>>>
>>>
>>>      
>>>
>>Okay, I buy it.
>>
>>a) Devs, who has a fcdsl, fcdsl2, fcdslsl, fcdslusb and/or fcdslslusb 
>>(AVM FRITZ!Card DSL)? Step up and take net-dialup/fcdsl under your wing!
>>b) Who has a device that works with net-dialup/hcfusbmodem and is 
>>willing to become its maintainer?
>>c)...
>>
>>In fact, who has any device that would work with some driver from 
>>net-dialup? I don't have _any_ device at all!
>>
>>    
>>
>
>Then how can we know anything in net-dialup actually works? Sounds like 
>net-dialup is sorely lacking devs with the relevant hardware.
>
>  
>
Well, judging after bugzilla, I would say that all my ebuilds works (or 
have worked when I've submitted them). When I've become member in 
net-dialup team, net-dialup had over 80 opened bugs, and this after I 
solved a few dozens in my trial period.
Does any dev has any complaints about my work? I think not.
You wanna increase "quality" of ebuilds by taking net-dialup from me? 
Okay, let the bugs pile up again!

Does any of the QA supporters looked at 
http://bugs.gentoo.org/reports.cgi?product=Gentoo+Linux&datasets=NEW%3A&datasets=ASSIGNED%3A&datasets=REOPENED%3A 
?
Gentoo isn't politics is getting the things done. At least this is my view.


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 256 bytes --]

Re: [gentoo-dev] Please follow keywording policy

[Ciaran McCreesh]

[-- Attachment #1: Type: text/plain, Size: 1009 bytes --]

On Thu, 10 Mar 2005 09:54:56 +0200 Alin Nastac <mrness@gentoo.org>
wrote:
| I always test unpack/compile/install functionality of the ebuild
| before  submittion, even if the submittion means only a script change.
| So, from my point of view, every ebuild of mine at least must get 
| installed. Of course, it doesn't mean that install process will
| succeed  on any other machines than mine...

What? "It compiles! Ship it!"?

| I'm only saying that no one will take your head if you do a mistake,
| but  when you're not doing nothing because you are fearing of a
| mistake -  this is problem. To err is human, you know.
| As for marking an ebuild stable over a month, this is the ideal case.
| On  average, I mark an ebuild stable on x86 after 2 months.

I find it rather worrying that this is considered acceptable on x86...

-- 
Ciaran McCreesh : Gentoo Developer (Vim, Fluxbox, shell tools)
Mail            : ciaranm at gentoo.org
Web             : http://dev.gentoo.org/~ciaranm


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

Re: [gentoo-dev] Please follow keywording policy

[Alin Nastac]

[-- Attachment #1: Type: text/plain, Size: 1649 bytes --]

Grant Goodyear wrote:

>Alin Nastac wrote: [Wed Mar 09 2005, 04:57:15PM CST]
>  
>
>>Btw, what is the sense of ~arch if not "testing"? No gentooer 
>>expects from a ~arch ebuild to be stable, so the sky would not fall if 
>>you made a mistake and release it under this keyword. When I hear "I 
>>cannot mark foo library as ~arch because I don't know how to test it" 
>>smells like excuse to me.
>>    
>>
>
>*Sigh*  The meaning of ~arch is that, at a minimum, the package works
>for the person who keyworded it (or, in some cases, worked for a trusted
>user on whose behalf the package was keyworded).   In other words, the
>dev believes that the package works, and that belief is based on
>evidence, not just wishful thinking.  An "arch" keyword means that there
>is considerable evidence that the package works for multiple people.
>Packages that might work, but also might not, ideally should not be in
>the tree at all, but could reasonably be package.mask'ed if testing is
>imminent.
>
>  
>
Not every time when I receive a new ebuild submittion, I also test that 
package because this is not always possible. Usually, I add the new 
ebuild with ~x86 and let testing to the user who request that. It may 
not be the orthodox way, but the risk of breaking something else in the 
process is 0 (a new ebuild means no other ebuilds depends on it).

Users don't usually come to me and say "that ebuild works for me". I 
take silence as a sign that everything works. I am sure I'm not the only 
one doing that.

I ask arches to mark a new ebuild as stable because a know bug have been 
solved or because the old stable version breaks something else.

[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 256 bytes --]

Re: [gentoo-dev] Please follow keywording policy

[Ciaran McCreesh]

[-- Attachment #1: Type: text/plain, Size: 454 bytes --]

On Thu, 10 Mar 2005 10:36:43 +0200 Alin Nastac <mrness@gentoo.org>
wrote:
| Users don't usually come to me and say "that ebuild works for me". I 
| take silence as a sign that everything works. I am sure I'm not the
| only  one doing that.

I hope you are the only one doing that...

-- 
Ciaran McCreesh : Gentoo Developer (Vim, Fluxbox, shell tools)
Mail            : ciaranm at gentoo.org
Web             : http://dev.gentoo.org/~ciaranm


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

Re: [gentoo-dev] Please follow keywording policy

[Chris Gianelloni]

[-- Attachment #1: Type: text/plain, Size: 3357 bytes --]

On Thu, 2005-03-10 at 00:57 +0200, Alin Nastac wrote: 
> I'm only annoyed by the bad attitude of some devs who will get involved 
> only what suits them, forgetting that if they would not help, no one 
> will. Btw, what is the sense of ~arch if not "testing"? No gentooer 
> expects from a ~arch ebuild to be stable, so the sky would not fall if 
> you made a mistake and release it under this keyword. When I hear "I 
> cannot mark foo library as ~arch because I don't know how to test it" 
> smells like excuse to me.

You didn't ask for ~arch, you asked for stable.  Also, testing doesn't
mean providing something that it broken.  The testing branch is for
*ebuild testing* not package testing.  A broken package should never be
added to the tree unless it is hard masked.  You're seeming very quick
to point to others and lay blame, but I just don't think you get the
fact that there are others who take QA of their systems more seriously
than you do.

> As for QA... does anyone think we *can* have proper QA procedures, with 
> our release speed and decentralized development model? And with only ... 
> 350 devs from which God knows how many are still active?  :-D
> Who thinks that clearly doesn't have a clue what QA means. It is 
> practically impossible to test every combination of ebuilds/USE/CFLAGS 
> so all we do is a surface test, letting the burden of testing on the 
> shoulders of our users.

This is true to an extent.  I know that I test ebuilds that I put into
portage with every combination of USE flags.  I also make sure that the
thing works.  I will also file bugs to myself, if need be, and ask users
for help with testing before putting something in the tree.  Being
tested does not necessarily mean that a developer did the testing, just
that a developer verified that it was tested.  If you aren't testing the
ebuilds you're committing, then you aren't doing QA and you're leaving
it up to others to discover if something you've added is broken.  This
should happen *before* it goes in the tree, not after.

> Despite of our unorthodox development process, many people believes 
> (including me) that our distro surclass traditional ones and is 
> generally more stable (go figure!).

There have been a few snafus here and there, but generally I would agree
with you.

> Maybe I'm too exigent, but I only ask from people to do what I do : be 
> genuinely interested in helping the devs who need it. Heck, I always try 
> to help any gentooer, dev or not. We all have our little systems because 
> our predecesors have worked on it, not because they  sit down and 
> debated whether to mark foo ebuild as ~arch or not.

Trying to force your ideas of QA onto another team isn't asking someone
to help you, it is asking someone to drop their beliefs in quality to
meet your timetables.  That is counter-productive more than helpful.
You're using a lot of emotional arguments, and none technical.

Could the mips team have helped quicker?  Sure.  Maybe it would have
been beneficial for them to have simply said, "Hey.  We don't have that
hardware so we can't test it." but trying to make them look like a bunch
of lazy developers isn't helping your case much... ;]

-- 
Chris Gianelloni
Release Engineering - Strategic Lead/QA Manager
Games - Developer
Gentoo Linux

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

Re: [gentoo-dev] Please follow keywording policy

[Chris Gianelloni]

[-- Attachment #1: Type: text/plain, Size: 587 bytes --]

On Thu, 2005-03-10 at 09:54 +0200, Alin Nastac wrote:
> Again, I want to emphasize that I didn't break anything in portage tree, 
> even if I have a "shoddy" attitude regarding Gentoo's QA.

Nobody has said that you did.  The complaint is that you're making an
entire team out to be bad guys because they didn't use your same methods
of "shoddy" QA.

If I were on an arch team, I wouldn't mark something I couldn't possibly
verify as working *at all* as stable, either.

-- 
Chris Gianelloni
Release Engineering - Strategic Lead/QA Manager
Games - Developer
Gentoo Linux

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

Re: [gentoo-dev] Please follow keywording policy

[Chris Gianelloni]

[-- Attachment #1: Type: text/plain, Size: 642 bytes --]

On Thu, 2005-03-10 at 08:27 +0000, Ciaran McCreesh wrote:
> | I'm only saying that no one will take your head if you do a mistake,
> | but  when you're not doing nothing because you are fearing of a
> | mistake -  this is problem. To err is human, you know.
> | As for marking an ebuild stable over a month, this is the ideal case.
> | On  average, I mark an ebuild stable on x86 after 2 months.
> 
> I find it rather worrying that this is considered acceptable on x86...

It isn't considered acceptable on any architecture.

-- 
Chris Gianelloni
Release Engineering - Strategic Lead/QA Manager
Games - Developer
Gentoo Linux

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

Re: [gentoo-dev] Please follow keywording policy

[Chris Gianelloni]

[-- Attachment #1: Type: text/plain, Size: 1741 bytes --]

On Thu, 2005-03-10 at 10:36 +0200, Alin Nastac wrote:
> Not every time when I receive a new ebuild submittion, I also test that 
> package because this is not always possible. Usually, I add the new 
> ebuild with ~x86 and let testing to the user who request that. It may 
> not be the orthodox way, but the risk of breaking something else in the 
> process is 0 (a new ebuild means no other ebuilds depends on it).

Add it as ~arch and p.mask it, or get more people to test it *while it
is still in bugzilla* until you are pretty sure that it works, *then*
add it.

If an ebuild has a DEPEND on >=foo-1.0 and you add foo-1.2, then it
*will* be pulled in as a dependency, so you can't possibly say that
nothing depends on it and be serious.

> Users don't usually come to me and say "that ebuild works for me". I 
> take silence as a sign that everything works. I am sure I'm not the only 
> one doing that.

No.  Everyone else is making sure the ebuild works before adding it.
The truth is that you've just been lucky, so far.

> I ask arches to mark a new ebuild as stable because a know bug have been 
> solved or because the old stable version breaks something else.

Great.  Don't start a hissy fit on -dev when they don't mark it stable
because they can't test it.  Especially when you haven't made a good
effort to contact them to resolve the problem.  Airing your dirty
laundry out in public just makes you look white trash... ;]  This isn't
Jerry Springer.  We don't need to know who your baby's daddy is.  A
simple email to the mips team could have kept all of this from even
being an issue.

-- 
Chris Gianelloni
Release Engineering - Strategic Lead/QA Manager
Games - Developer
Gentoo Linux

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

Re: [gentoo-dev] Please follow keywording policy

[Alin Nastac]

[-- Attachment #1: Type: text/plain, Size: 2764 bytes --]

Chris Gianelloni wrote:

>On Thu, 2005-03-10 at 10:36 +0200, Alin Nastac wrote:
>  
>
>>Not every time when I receive a new ebuild submittion, I also test that 
>>package because this is not always possible. Usually, I add the new 
>>ebuild with ~x86 and let testing to the user who request that. It may 
>>not be the orthodox way, but the risk of breaking something else in the 
>>process is 0 (a new ebuild means no other ebuilds depends on it).
>>    
>>
>
>Add it as ~arch and p.mask it, or get more people to test it *while it
>is still in bugzilla* until you are pretty sure that it works, *then*
>add it.
>
>If an ebuild has a DEPEND on >=foo-1.0 and you add foo-1.2, then it
>*will* be pulled in as a dependency, so you can't possibly say that
>nothing depends on it and be serious.
>
>  
>
Please read what I've wrote above. I said "I receive a new ebuild", 
wouldn't I?
You can keep your straight face...

As for ~arch and p.mask, please read again 
http://www.gentoo.org/proj/en/devrel/handbook/handbook.xml?part=3&chap=1#doc_chap4 
under "Masked packages".

>>Users don't usually come to me and say "that ebuild works for me". I 
>>take silence as a sign that everything works. I am sure I'm not the only 
>>one doing that.
>>    
>>
>
>No.  Everyone else is making sure the ebuild works before adding it.
>The truth is that you've just been lucky, so far.
>
>  
>
Yeah, luck, that's for sure! I would be more careful before I would make 
such implausible statements. I did worked in portage for about 5-6 
months, you know, and I wasn't idling on IRC! What are the odds to keep 
being lucky every time ?

>>I ask arches to mark a new ebuild as stable because a know bug have been 
>>solved or because the old stable version breaks something else.
>>    
>>
>
>Great.  Don't start a hissy fit on -dev when they don't mark it stable
>because they can't test it.  Especially when you haven't made a good
>effort to contact them to resolve the problem.  Airing your dirty
>laundry out in public just makes you look white trash... ;]  This isn't
>Jerry Springer.  We don't need to know who your baby's daddy is.  A
>simple email to the mips team could have kept all of this from even
>being an issue.
>
>  
>
I did apologized to mips team, remember? Again, I apologies for my 
overreaction. I should have waited for a month...

This disscution is started to be both juvenile and counter-productive. I 
regret that.

Btw, what is your position, being QA manager and all, regarding 
http://bugs.gentoo.org/reports.cgi?product=Gentoo+Linux&datasets=NEW%3A&datasets=ASSIGNED%3A&datasets=REOPENED%3A 
?
Isn't it one of the top QA's priorities to assure that all known bugs 
are resolved? Or, as ciaranm's membership to mips, you don't do that 
part of the QA?


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 256 bytes --]

Re: [gentoo-dev] Please follow keywording policy

[Chris Gianelloni]

[-- Attachment #1: Type: text/plain, Size: 3184 bytes --]

On Thu, 2005-03-10 at 21:00 +0200, Alin Nastac wrote:
> Please read what I've wrote above. I said "I receive a new ebuild", 
> wouldn't I?
> You can keep your straight face...

What?

That makes absolutely no sense, so I really can't say anything on it.

> As for ~arch and p.mask, please read again 
> http://www.gentoo.org/proj/en/devrel/handbook/handbook.xml?part=3&chap=1#doc_chap4 
> under "Masked packages".

Alright.

"Package.mask is used to prevent merging of packages that are broken,
break something else, or badly need testing before going into ~ARCH
KEYWORDS in the tree."

I would say a package that you've not even tested for functionality
falls under "badly need testing before going into ~ARCH" pretty
squarely.

Now, let's look under the "~ARCH in KEYWORDS" on that same document.

"There is a difference between using package.mask and ~arch for ebuilds.
The use of ~arch denotes an ebuild requires testing."

The word ebuild is even in bold.

Did you even bother reading this before trying to use it to make a
point?

> Yeah, luck, that's for sure! I would be more careful before I would make 
> such implausible statements. I did worked in portage for about 5-6 
> months, you know, and I wasn't idling on IRC! What are the odds to keep 
> being lucky every time ?

Pretty high, apparently.

You are relying 100% on other people to do your testing for you.
Whether it is the upstream authors or the users of the package.

> This disscution is started to be both juvenile and counter-productive. I 
> regret that.

I'm not sure how, but if you're regretting your statements, perhaps you
should put more thought into them before shouting "Fire!" in a crowded
theatre.

> Btw, what is your position, being QA manager and all, regarding 
> http://bugs.gentoo.org/reports.cgi?product=Gentoo+Linux&datasets=NEW%3A&datasets=ASSIGNED%3A&datasets=REOPENED%3A 
> ?

For one, you are now trying to point at *ME* for something?  I'm sorry,
but I don't have to justify myself to you.  I am doing my job properly.

Also, I have nothing to do with the QA project.  I am the QA Manager for
Release Engineering.

Perhaps you should spend all this energy testing your own packages,
rather than trying to point out where you think others are doing wrong
when you don't know what you're talking about.

Also, let's look at that graph.  I would be worried if the number of
bugs that were at REOPEN were growing that fast, but considering the
number of packages being added to the portage tree since October of
2003, I don't see that as being too odd.

> Isn't it one of the top QA's priorities to assure that all known bugs 
> are resolved? Or, as ciaranm's membership to mips, you don't do that 
> part of the QA?

Like I said, you really need to back off.  At this point, you're trying
to pick a fight with me simply because you have no retort.  I'm going to
request this one time from you.  Stop.

There is no need for you to stoop to such immature tactics, and it won't
be tolerated.

Thanks,

-- 
Chris Gianelloni
Release Engineering - Strategic Lead/QA Manager
Games - Developer
Gentoo Linux

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

Re: [gentoo-dev] Please follow keywording policy

[John Myers]

[-- Attachment #1: Type: text/plain, Size: 721 bytes --]

On Thursday 10 March 2005 11:33, Chris Gianelloni wrote:
> On Thu, 2005-03-10 at 21:00 +0200, Alin Nastac wrote:
> > Please read what I've wrote above. I said "I receive a new ebuild",
> > wouldn't I?
> > You can keep your straight face...
>
> What?
>
> That makes absolutely no sense, so I really can't say anything on it.
>
I think what he means to say is "I receive an ebuild for a new package", 
rather than "I receive an ebuild for a new version of an existing package". 
If I am correct, what he said makes some sense.

-- 
t3h 3l3ctr0n3rd <electronerd@monolith3d.com>
Supermarket Deli Clerk and Student Programmer

OpenPGP Key Fingerprint:
    0A65 EEFA B23A F0AC E6C2 C71C BEA0 E055 BE0E EC25

[-- Attachment #2: Type: application/pgp-signature, Size: 198 bytes --]