Mailing List Archive

Danny van Dyk schrieb:
> Am Samstag, 2. September 2006 13:18 schrieb Edgar Hucek:
>>>> 2.) Enable the use flage accessibility gnome cant be
>>>> merged. It fails on compile the speech-tools.
>>>> It seams that USE flags are not realy tested or how
>>>> can it happen that there are already know bugs in the
>>>> stable distro ?
>>>>
>>>> http://bugs.gentoo.org/show_bug.cgi?id=116030
>>>>
>>>> Festival and the speech-tools are well know not to
>>>> compile with gcc >=4.
>>> Well, you know - if you go to read the speech-tools/festival & co.
>>> bug, and read the ebuild, you'll see that the whole thing and code
>>> is one huge mess, that doesn't compile even w/ gcc-3.3 without
>>> patching. You'd probably prefer to never put out a new release, I
>>> guess? How many people are using this one, and how does it justify
>>> delaying the release even more?
>> From my point of view, should it be garanted that a package and
>> depencies compiles when all use flags are enabled. If a depency can't
>> be compiled the use flag and depence should be dissabled/removed from
>> a package.
> Please _think_ before you make such a demand. Just a small investigation
> would show this:
>
> dev-lang/php-5.1.4-r6 has _96_ USE flags. That makes 2^96 = 7.9928+28
> combinations. Given the (unreasonable) assumption that each compilation
> would only take 1s and each compilation would actually succeed, you'd
> still have ~8e28 seconds. The age of the universe is approximately 4e17
> seconds.
>
> This hasn't yet investigated allt he possible combinations of packages
> depending on dev-lang/php, or the ~10,000 other packages in the tree.
>
> Danny

Just a side hint. Try to enable all flags at the first cimpile time would
reduce trys drasticaly ;)
So you say a developer cant't test all useflags? That is a strange
message from you. How can a developer garantee that his package is correct.
Realy funny, i only hear exuses but no real solution for the problem.
The fact is, that long outstanding bugs are simple ignored. If a useflag
would only apply to one package it could be ok, but not when the same
useflag is in other packages and makes this one useflag for the "normal user"
unusable.

cu

Edgar (gimli) Hucek

--
gentoo-dev@gentoo.org mailing list

On 02/09/06, Edgar Hucek <gimli@gentoo.org> wrote:

> Realy funny, i only hear exuses but no real solution for the problem.


The universe ending before testing is finished is a pretty good excuse.

Edgar Hucek wrote:
> Just a side hint. Try to enable all flags at the first cimpile time would
> reduce trys drasticaly ;)

If you had a look at the php ebuild (just because we took it as example
here), you'd see that it is a bit more complicated than just enabling
everything to have everything tested.

> So you say a developer cant't test all useflags? That is a strange
> message from you. How can a developer garantee that his package is correct.

He can't. That's what we're saying. Nobody said we can, nor do, nor want to.

> Realy funny, i only hear exuses but no real solution for the problem.

You have heard the real solution for the specific problems you pointed
out: File a bug. You have also heard why it is impossible to guarantee
that it simply works.

> The fact is, that long outstanding bugs are simple ignored. If a useflag
> would only apply to one package it could be ok, but not when the same
> useflag is in other packages and makes this one useflag for the "normal user"
> unusable.

man portage:

package.use
Per-package USE flags. Useful for tracking local USE
flags or for enabling USE flags for certain packages
only. Perhaps you develop GTK and thus you want documen-
tation for it, but you don't want documentation for QT.
Easy as pie my friend!

Format:
- comments begin with #
- one DEPEND atom per line with space-delimited USE flags

Example:
# turn on docs for GTK 2.x
=x11-libs/gtk+-2* doc
# disable mysql support for QT
x11-libs/qt -mysql

Know your tools, man.

--
Kind Regards,

Simon Stelling
Gentoo/AMD64 developer
--
gentoo-dev@gentoo.org mailing list

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

Edgar Hucek wrote:
> Danny van Dyk schrieb:
>> Am Samstag, 2. September 2006 13:18 schrieb Edgar Hucek:
>>>>> 2.) Enable the use flage accessibility gnome cant be
>>>>> merged. It fails on compile the speech-tools.
>>>>> It seams that USE flags are not realy tested or how
>>>>> can it happen that there are already know bugs in the
>>>>> stable distro ?
>>>>>
>>>>> http://bugs.gentoo.org/show_bug.cgi?id=116030
>>>>>
>>>>> Festival and the speech-tools are well know not to
>>>>> compile with gcc >=4.
>>>> Well, you know - if you go to read the speech-tools/festival & co.
>>>> bug, and read the ebuild, you'll see that the whole thing and code
>>>> is one huge mess, that doesn't compile even w/ gcc-3.3 without
>>>> patching. You'd probably prefer to never put out a new release, I
>>>> guess? How many people are using this one, and how does it justify
>>>> delaying the release even more?
>>> From my point of view, should it be garanted that a package and
>>> depencies compiles when all use flags are enabled. If a depency can't
>>> be compiled the use flag and depence should be dissabled/removed from
>>> a package.
>> Please _think_ before you make such a demand. Just a small investigation
>> would show this:
>>
>> dev-lang/php-5.1.4-r6 has _96_ USE flags. That makes 2^96 = 7.9928+28
>> combinations. Given the (unreasonable) assumption that each compilation
>> would only take 1s and each compilation would actually succeed, you'd
>> still have ~8e28 seconds. The age of the universe is approximately 4e17
>> seconds.
>>
>> This hasn't yet investigated allt he possible combinations of packages
>> depending on dev-lang/php, or the ~10,000 other packages in the tree.
>>
>> Danny
>
> Just a side hint. Try to enable all flags at the first cimpile time would
> reduce trys drasticaly ;)
> So you say a developer cant't test all useflags? That is a strange
> message from you. How can a developer garantee that his package is correct.
> Realy funny, i only hear exuses but no real solution for the problem.
> The fact is, that long outstanding bugs are simple ignored. If a useflag
> would only apply to one package it could be ok, but not when the same
> useflag is in other packages and makes this one useflag for the "normal user"
> unusable.
>
> cu
>
> Edgar (gimli) Hucek
>
Edgar-

You clearly have absolutely no idea how development and testing happens.
This is *free* software with no warranty. Our releases are tested with
the profile defaults provided in the release. Nothing more. If that's
not good enough for you, please find a distribution that you have to pay
for like RHEL. Their testing is no better than ours, but at least
paying something entitles you to bitch at them.

- --
=======================================================
Mike Doty kingtaco -at- gentoo.org
Gentoo/AMD64 Strategic Lead
Gentoo Developer Relations
Gentoo Recruitment Lead
Gentoo Infrastructure
GPG: E1A5 1C9C 93FE F430 C1D6 F2AF 806B A2E4 19F4 AE05
=======================================================
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.5 (GNU/Linux)

iQCVAwUBRPmNU4BrouQZ9K4FAQLu0QQAwHVnw/zCbHTjDLb3h50tMiUkdgfhZTpF
sEYpsee/LlgYpoVqZoukOQ7X3h8N5uRaHNU/SkcS6blMYGNGhdbPuu9taOylp+x1
6BoXi7FlA3tbSpmRQQdsSO3/fqWwS26lHYKtvkYkhFfqjSP+wd3NZPBlUH4hVpbo
id2I+hvq8R0=
=3jiw
-----END PGP SIGNATURE-----
--
gentoo-dev@gentoo.org mailing list

Simon Stelling schrieb:
> Edgar Hucek wrote:
>> Just a side hint. Try to enable all flags at the first cimpile time would
>> reduce trys drasticaly ;)
>
> If you had a look at the php ebuild (just because we took it as example
> here), you'd see that it is a bit more complicated than just enabling
> everything to have everything tested.
>
>> So you say a developer cant't test all useflags? That is a strange
>> message from you. How can a developer garantee that his package is correct.
>
> He can't. That's what we're saying. Nobody said we can, nor do, nor want to.
>
>> Realy funny, i only hear exuses but no real solution for the problem.
>
> You have heard the real solution for the specific problems you pointed
> out: File a bug. You have also heard why it is impossible to guarantee
> that it simply works.
>
>> The fact is, that long outstanding bugs are simple ignored. If a useflag
>> would only apply to one package it could be ok, but not when the same
>> useflag is in other packages and makes this one useflag for the "normal user"
>> unusable.
>
> man portage:
>
> package.use
> Per-package USE flags. Useful for tracking local USE
> flags or for enabling USE flags for certain packages
> only. Perhaps you develop GTK and thus you want documen-
> tation for it, but you don't want documentation for QT.
> Easy as pie my friend!
>
> Format:
> - comments begin with #
> - one DEPEND atom per line with space-delimited USE flags
>
> Example:
> # turn on docs for GTK 2.x
> =x11-libs/gtk+-2* doc
> # disable mysql support for QT
> x11-libs/qt -mysql
>
> Know your tools, man.
>

I know my tools but not necessarly the normal user who wanna use gentoo
and is ending frustrated.

cu

Edgar (gimli) Hucek
--
gentoo-dev@gentoo.org mailing list

Edgar Hucek wrote:
> Danny van Dyk schrieb:
>> This hasn't yet investigated allt he possible combinations of packages
>> depending on dev-lang/php, or the ~10,000 other packages in the tree.
>>
>> Danny
>
> Just a side hint. Try to enable all flags at the first cimpile time would
> reduce trys drasticaly ;)

Yes, it would indeed drastically reduce the time to almost zero due to
use flag collisions... :)

> So you say a developer cant't test all useflags? That is a strange
> message from you.

No, even PHP devs can't test them all, and definitely not all their
combinations (simple maths, see previous mail). Not to mention that some
of the flags require commercial software installed that's not in
portage, so they are actually unsupported.


--
Best regards,

Jakub Moc
mailto:jakub@gentoo.org
GPG signature:
http://subkeys.pgp.net:11371/pks/lookup?op=get&search=0xCEBA3D9E
Primary key fingerprint: D2D7 933C 9BA1 C95B 2C95 B30F 8717 D5FD CEBA 3D9E

... still no signature ;)

Edgar Hucek wrote:
> I know my tools but not necessarly the normal user who wanna use gentoo
> and is ending frustrated.

If the users are too lazy to read the documentation, why should we care
about them?

--
Kind Regards,

Simon Stelling
Gentoo/AMD64 developer
--
gentoo-dev@gentoo.org mailing list

> I know my tools but not necessarly the normal user who wanna use gentoo
> and is ending frustrated.
>
> cu
>
> Edgar (gimli) Hucek

Enrico? Is that you in disguise?

--
gentoo-dev@gentoo.org mailing list

Simon Stelling <blubb@gentoo.org> posted 44F991D7.9060905@gentoo.org,
excerpted below, on Sat, 02 Sep 2006 16:14:47 +0200:

> Edgar Hucek wrote:
>> I know my tools but not necessarly the normal user who wanna use gentoo
>> and is ending frustrated.
>
> If the users are too lazy to read the documentation, why should we care
> about them?

Exactly.

To such a "normal user", unwilling to invest the very real time and energy
into learning about Gentoo and how to customize it to his wishes, most if
not all Gentoo devs will be happy to recommend Ubuntu or whatever. Ubuntu
is by all reports a very respectable distribution, arguably one of the
most user friendly yet powerful out there. (Linspire/Freespire's probably
the most user friendly, disregarding power.)

Let Gentoo do what Gentoo does best, cater to those that /like/ that
customizability, even, perhaps /because/ of, the challenge of mastering
the machine and bending it to our will. There are plenty of other
distributions out there for those that are more interested in just having
it work, with as little knowledge and effort invested on their part as
possible.

--
Duncan - List replies preferred. No HTML msgs.
"Every nonfree program has a lord, a master --
and if you use the program, he is your master." Richard Stallman

--
gentoo-dev@gentoo.org mailing list

Seems my message got swallowed...

On Saturday 02 September 2006 15:36, Edgar Hucek wrote:
> Just a side hint. Try to enable all flags at the first cimpile time would
> reduce trys drasticaly ;)

There are lots of use flag combinations incompatible with each other within a
package as well as packages relying on other ones to be build with or without
use flags of other packages. The number of pssoble combinations would is too
high, even if we had build servers running around the clock.

In case of point two, you're right, that it doesn't let Gentoo look good.
Neither Gnome nor KDE (no use flag in this case) accessibiliy stuff builds
now - and bug 116030 is open since nine months. Partly the problem is that
we're understaffed, partly - and this is my very personal opinion - the
problem is that releasing with GCC 4.x has been rushed - speak the notice
came one or two months too late.


Carsten

On Sunday 03 September 2006 00:16, Carsten Lohrke wrote:
> Neither Gnome nor KDE (no use flag in this case) accessibiliy stuff builds
> now - and bug 116030 is open since nine months.
And waiting other 2, 3, 4, 5, 6 months won't change the thing. Why? Because we
have _no_ accessibility team right now. If we had one, the problem would have
been solved. Unfortunately that software is doomed to lag behind the rest of
Gentoo unless someone maintain it. If it wasn't for the need of that software
by some users, probably treecleaners would have removed that already.

In _this_ particular case, the notice interval is not important.

--
Diego "Flameeyes" Pettenò - http://farragut.flameeyes.is-a-geek.org/
Gentoo/Alt lead, Gentoo/FreeBSD, Video, AMD64, Sound, PAM, KDE

On Sat, Sep 02, 2006 at 08:55:33AM -0500, Mike Doty wrote:
> If that's not good enough for you, please find a distribution that you
> have to pay for like RHEL. Their testing is no better than ours, but
> at least paying something entitles you to bitch at them.
Or consider paying a Gentoo developer [*] as your first level support
person, and liaison with Gentoo. Thus they consult for you, and tell you
if your specific combinations are going to work, and do their hardest to
keep them working.

But don't forget to heed their warnings of what you should and shouldn't
do.

I believe there are several developers that are unemployed, and would
like more work of this nature.

* I'm aware that myself and several other developers do this in various
ways, most commonly by having our regular employer task us with making
sure that changes in Gentoo won't break what we do. However I'm not
taking any new consulting clients presently.

--
Robin Hugh Johnson
E-Mail : robbat2@gentoo.org
GnuPG FP : 11AC BA4F 4778 E3F6 E4ED F38E B27B 944E 3488 4E85

Simon Stelling wrote:

> Edgar Hucek wrote:
> > I know my tools but not necessarly the normal user who wanna use gentoo
> > and is ending frustrated.
>
> If the users are too lazy to read the documentation, why should we care
> about them?

Because we risk that Gentoo may receive the "user-UN-friendly" label and
become irrelevant in the long run? I know it ain't gonna happen, but still.

Both Edgar and you have some valid points. He refers mostly to the out-of-box
experience, which includes compiling GNOME and its dependencies at the install
time. With USE="accessibility" enabled, which makes perfect sense for people
with disabilities. And then the first-ever Gentoo installation breaks on the
speech-tools and festival.

How would *you* feel in such case?

You OTOH bring to the table a fact that developers shouldn't be that much
concerned with the stabilization/testing of packages before new release of
installation media. But new releases *ARE* targeted specifically at new users
and it's them who suffer the most. Next to it is the reputation of Gentoo and
its developers. Edgar's call was targeted mostly at releng and QA teams, who
should poke developers to decrease number of similar problems.

I maintain a bunch of Debian/sparc, Debian/i386, Gentoo/amd64, Gentoo/x86,
Solaris/sparc, Ubuntu/i686 boxes and mind you, out-of-box experience at
install time means A LOT.

More respect to the users => more respect to Gentoo.

Regards,
Wiktor Wandachowicz

PS. I'm already on the CC list of bug #116030 for the same reasons, but
I've been mostly quiet because I do know my tools ;) But OTOH I've been
already running Gentoo for a while....

--
gentoo-dev@gentoo.org mailing list

On Sunday 03 September 2006 00:42, Diego 'Flameeyes' Pettenò wrote:
> And waiting other 2, 3, 4, 5, 6 months won't change the thing. Why? Because
> we have _no_ accessibility team right now.

Well, the bug is assigned to williamh, who is not /completely/ inactive. I
wonder, if only 37 commits in more than two years suffices for cvs access,
though.

> If we had one, the problem would
> have been solved. Unfortunately that software is doomed to lag behind the
> rest of Gentoo unless someone maintain it. If it wasn't for the need of
> that software by some users, probably treecleaners would have removed that
> already.
>
> In _this_ particular case, the notice interval is not important.

You're wrong here. What I'm inclined about is that we had (leastwise) a
fourteen day short notice to when the releaase snapshot would be taken. To
the end of this time frame there was another one that we'd release with GCC
4.x. Even if we had enough people to deal with everything thrown at us,it
would have been impossible to fix and stabilize the relevant packages on all
architectures.

If I had known this as estimated goal two months earlier, I'd had switched to
GCC 4.x a while before and noticed the bug, instead when it is too late. I
consider this part of what is broken within Gentoo communication-wise.


Carsten

Wiktor Wandachowicz wrote:
> Simon Stelling wrote:
>
>> Edgar Hucek wrote:
>>> I know my tools but not necessarly the normal user who wanna use gentoo
>>> and is ending frustrated.
>> If the users are too lazy to read the documentation, why should we care
>> about them?
>
> Because we risk that Gentoo may receive the "user-UN-friendly" label and
> become irrelevant in the long run? I know it ain't gonna happen, but still.
>
> Both Edgar and you have some valid points. He refers mostly to the out-of-box
> experience, which includes compiling GNOME and its dependencies at the install
> time. With USE="accessibility" enabled, which makes perfect sense for people
> with disabilities. And then the first-ever Gentoo installation breaks on the
> speech-tools and festival.
>
> How would *you* feel in such case?
>
> You OTOH bring to the table a fact that developers shouldn't be that much
> concerned with the stabilization/testing of packages before new release of
> installation media. But new releases *ARE* targeted specifically at new users
> and it's them who suffer the most. Next to it is the reputation of Gentoo and
> its developers. Edgar's call was targeted mostly at releng and QA teams, who
> should poke developers to decrease number of similar problems.
>
> I maintain a bunch of Debian/sparc, Debian/i386, Gentoo/amd64, Gentoo/x86,
> Solaris/sparc, Ubuntu/i686 boxes and mind you, out-of-box experience at
> install time means A LOT.
>
> More respect to the users => more respect to Gentoo.
>

Let's see...

Several points (misunderstandings) need to be clarified.

1) Gentoo is not intended to be an out-of-the-box distro, but instead, a
customizable distro. Can see the difference?.. There are many, one of
them is that users should 'make' the process of using Gentoo _friendly_
partially by themselves through reading documentation and tutorial when
needed (and sometimes going through a list of bugs to know what it is
going on).

2) Gentoo releases are "very".touppercase different to most of the other
distros. Gentoo releases are mainly intended to be used as a tool to get
you started building your _own_ system in an automatic way through
scripts/metadata, this being very different to other distros, where they
simply force you to use version 6.6.6 as a bunch of dead packages that
won't likely suffer any major changes within the next 6 months until
upgrading (which can be a very painful process) to the next 6.6.7 release.

This is precisely why i say Gentoo is an incremental meta-distro.

3) Considering the two points above, i therefore think , there is no
point (and actually makes no sense) to bitch at our releng team (which
did a great job) because two packages don't currently compile.

4) Gentoo is more a community than anything else. So we indeed all
deserve respect. Some developers put into this project (the releng team
being one of them) a lot of effort, so making comments like this thread
might be very insulting for many people; apart of making false claims
that could lead to a bunch of misconceptions. Now who is being
disrespectful?

If neither of those points are convincing enough, then remember free
software comes with *NO-WARRANTY*

Thanks,

My 0.2bs



--


Luis F. Araujo "araujo at gentoo.org"
Gentoo Linux


--
gentoo-dev@gentoo.org mailing list

Luis Francisco Araujo wrote:
>
> If neither of those points are convincing enough, then remember free
> software comes with *NO-WARRANTY*

s/free//

Even payware is w/out warranties.

lu

--

Luca Barbato

Gentoo/linux Gentoo/PPC
http://dev.gentoo.org/~lu_zero

--
gentoo-dev@gentoo.org mailing list

Wiktor Wandachowicz wrote:
>> If the users are too lazy to read the documentation, why should we care
>> about them?
>
> Because we risk that Gentoo may receive the "user-UN-friendly" label and
> become irrelevant in the long run? I know it ain't gonna happen, but still.

Well, that might be the case, but then, do we really care? I'm more than
glad if people who are too lazy to read the docs don't bugger me with
their issues which they could easily avoid if they'd have read the docs.

> You OTOH bring to the table a fact that developers shouldn't be that much
> concerned with the stabilization/testing of packages before new release of
> installation media. But new releases *ARE* targeted specifically at new users
> and it's them who suffer the most. Next to it is the reputation of Gentoo and
> its developers. Edgar's call was targeted mostly at releng and QA teams, who
> should poke developers to decrease number of similar problems.

I never said that, and I don't mean it either. We are concerned about
testing, but we can only do as much as human can do. Gentoo gives you
much flexibility. We can workaround the issue by masking the
accessibility flag for example, but that wouldn't be all that great,
because it only hides the problem, and it doesn't help those disabled
people either.

> I maintain a bunch of Debian/sparc, Debian/i386, Gentoo/amd64, Gentoo/x86,
> Solaris/sparc, Ubuntu/i686 boxes and mind you, out-of-box experience at
> install time means A LOT.

I know that. The first Gentoo CD I threw away just after booting it
because I couldn't figure out how to launch the setup app. "Wow, what a
crappy shit." I thought. Seriously, I don't want such people to use the
distro. It's not the right one for them.

(On a sidenote, should be quite obvious that in a second try I did
figure out how to install Gentoo and kind of changed my mind ;))

> More respect to the users => more respect to Gentoo.

I'm not sure how to parse that, but in case you mean that Gentoo gets
more popular when we make the out-of-the-box-experience better, then I
agree. I do not agree that this is something I want, though, because to
achieve that you either need a) much more resources or b) drop some of
the flexibility we offer. a) is hard to get and b) sucks. I'd rather
have other people think Gentoo is a bad distro but be happy with it
myself. Yes, I am a selfish pig.

--
Kind Regards,

Simon Stelling
Gentoo/AMD64 developer
--
gentoo-dev@gentoo.org mailing list

Either MTA or MUA brokeness. Another email I have to send a second time. :(

On Sunday 03 September 2006 00:42, Diego 'Flameeyes' Pettenò wrote:
> And waiting other 2, 3, 4, 5, 6 months won't change the thing. Why? Because
> we have _no_ accessibility team right now.

Well, the bug is assigned to williamh, who is not /completely/ inactive. I
wonder, if only 37 commits in more than two years suffices for cvs access,
though.

> If we had one, the problem would
> have been solved. Unfortunately that software is doomed to lag behind the
> rest of Gentoo unless someone maintain it. If it wasn't for the need of
> that software by some users, probably treecleaners would have removed that
> already.
>
> In _this_ particular case, the notice interval is not important.

You're wrong here. What I'm inclined about is that we had (leastwise) a
fourteen day short notice to when the releaase snapshot would be taken. To
the end of this time frame there was another one that we'd release with GCC
4.x. Even if we had enough people to deal with everything thrown at us,it
would have been impossible to fix and stabilize the relevant packages on all
architectures.

If I had known this as estimated goal two months earlier, I'd had switched to
GCC 4.x a while before and noticed the bug, instead when it is too late. I
consider this part of what is broken within Gentoo communication-wise.


Carsten

On Sunday 03 September 2006 15:02, Carsten Lohrke wrote:
> You're wrong here. What I'm inclined about is that we had (leastwise) a
> fourteen day short notice to when the releaase snapshot would be taken. To
> the end of this time frame there was another one that we'd release with GCC
> 4.x. Even if we had enough people to deal with everything thrown at us,it
> would have been impossible to fix and stabilize the relevant packages on
> all architectures.
Read me well please, I mean that in the case of the accessibility packages
here brought to our attention, the interval is irrelevant.

I never said that it wouldn't have helped to have it longer for many other
little things, of course, and in that I agree with you.

--
Diego "Flameeyes" Pettenò - http://farragut.flameeyes.is-a-geek.org/
Gentoo/Alt lead, Gentoo/FreeBSD, Video, AMD64, Sound, PAM, KDE

Le Sun, 03 Sep 2006 07:29:47 -0400,
Luis Francisco Araujo <araujo@gentoo.org> a écrit :

> Wiktor Wandachowicz wrote:
> > Simon Stelling wrote:
> >
> >> Edgar Hucek wrote:
> >>> I know my tools but not necessarly the normal user who wanna use gentoo
> >>> and is ending frustrated.
> >> If the users are too lazy to read the documentation, why should we care
> >> about them?
> >
> > Because we risk that Gentoo may receive the "user-UN-friendly" label and
> > become irrelevant in the long run? I know it ain't gonna happen, but still.
> >
> > Both Edgar and you have some valid points. He refers mostly to the
> > out-of-box experience, which includes compiling GNOME and its dependencies
> > at the install time. With USE="accessibility" enabled, which makes perfect
> > sense for people with disabilities. And then the first-ever Gentoo
> > installation breaks on the speech-tools and festival.
> >
> > How would *you* feel in such case?
> >
> > You OTOH bring to the table a fact that developers shouldn't be that much
> > concerned with the stabilization/testing of packages before new release of
> > installation media. But new releases *ARE* targeted specifically at new
> > users and it's them who suffer the most. Next to it is the reputation of
> > Gentoo and its developers. Edgar's call was targeted mostly at releng and
> > QA teams, who should poke developers to decrease number of similar problems.
> >
> > I maintain a bunch of Debian/sparc, Debian/i386, Gentoo/amd64, Gentoo/x86,
> > Solaris/sparc, Ubuntu/i686 boxes and mind you, out-of-box experience at
> > install time means A LOT.
> >
> > More respect to the users => more respect to Gentoo.
> >
>
> Let's see...
>
> Several points (misunderstandings) need to be clarified.
>
> 1) Gentoo is not intended to be an out-of-the-box distro, but instead, a
> customizable distro. Can see the difference?.. There are many, one of
> them is that users should 'make' the process of using Gentoo _friendly_
> partially by themselves through reading documentation and tutorial when
> needed (and sometimes going through a list of bugs to know what it is
> going on).
>
It a big difference between USE flags and compilation flags. It is unsafe
compilation flags and the doc recommend the use of safe CFLAGS, when nowhere
the doc say: Don't use this USE flag, it is unsafe or can break things.!

The fault is not with portage, it is even a better package management system as
apt-get (it is my main reason to use gentoo, it take time to install, but I
can do something else in the meantime, when I can do nothing else when I
backup and reinstall years of work with another distro when doing a major
upgrade).

The fault is not in the doc either: a USE flag must be safe and just work. I
understand at some few flags will not work in all cases, as example an ati USE
flag will not work with a nvidia card, but a flag as accessibility must work in
all cases.

So the fault can be programmers or the system. I don't know enough at that time
about the framework within gentoo programmers are working, but maybe at it is a
combination of the 2.

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

On Sat, 2006-09-02 at 23:11 -0600, Ryan Hill wrote:
> Carsten Lohrke wrote:
>
> > we're understaffed, partly - and this is my very personal opinion - the
> > problem is that releasing with GCC 4.x has been rushed
>
> I'd have to agree with you on that. I understand the appeal of exciting
> press releases but there were over 75 GCC 4.1 bugs still open for
> problems in *~arch* when the decision was made to go stable. Even now
> there's more than 50 left, with an equal and growing number of stable bugs.

It had nothing to do with press releases and more to do with the fact
that 50 or even 75 out of > 10,000 is beans. Also, most of that stuff
is in packages that are either unmaintained, or the maintainers are
focusing efforts in other places. Having a bug report open for 2 months
*should* be *plenty* of time for maintainers to fix their packages.

Here's the thing, there are certain points when you just have to say
that you're not waiting on a few people who aren't keeping up anymore.
Would you rather us lower the standards (just like the US education
system) for the few that are a bit slow?

--
Chris Gianelloni
Release Engineering - Strategic Lead
x86 Architecture Team
Games - Developer
Gentoo Linux

On Sun, 2006-09-03 at 07:41 +0000, Wiktor Wandachowicz wrote:
> its developers. Edgar's call was targeted mostly at releng and QA teams, who
> should poke developers to decrease number of similar problems.

Sorry, but Release Engineering has no wishes to become the "Gentoo
Developer Babysitting Project" at this time. We would much prefer work
on our release media. If some project, for any reason, is not up to
snuff, it is *not* our job to fix it. There are simply too many
projects out there. Would you rather we switch to a Debian model where
everything has to be perfect, but we don't release for 5 years?

--
Chris Gianelloni
Release Engineering - Strategic Lead
x86 Architecture Team
Games - Developer
Gentoo Linux

Chris Gianelloni wrote:
> On Sun, 2006-09-03 at 07:41 +0000, Wiktor Wandachowicz wrote:
>> its developers. Edgar's call was targeted mostly at releng and QA teams, who
>> should poke developers to decrease number of similar problems.
>
> Sorry, but Release Engineering has no wishes to become the "Gentoo
> Developer Babysitting Project" at this time. We would much prefer work
> on our release media. If some project, for any reason, is not up to
> snuff, it is *not* our job to fix it. There are simply too many
> projects out there. Would you rather we switch to a Debian model where
> everything has to be perfect, but we don't release for 5 years?
>

I'd rather have people ignore trolls (hi Enrico in disguise)

If something is broken and _NOBODY_ noticed it before either:
- it isn't a showstopper for most of the devs and we hadn't got the
complaints from our users
- it should be p.masked or updated with medium priority.

if our favourite cu-troll wants to point that some programs written by a
dog are ugly he could use the same time to fill a proper bug and or
provide patches.

that said releng has the duty to just provide a livecd that works and
stages that could be used to start getting a working system and I think
they succeeded as usual.

There are some known issue pointed already but nothing could be perfect.

"Release quickly & release often" maybe isn't really THE solution, but
"make it to the deadline decently well and plan a -r1 to address some of
the known issues that could hit more people in the future(eg new hw
support)" looks good enough.

that said I hope that the sales on trollfood will end ...

--

Luca Barbato

Gentoo/linux Gentoo/PPC
http://dev.gentoo.org/~lu_zero

--
gentoo-dev@gentoo.org mailing list

Carsten Lohrke wrote:

> we're understaffed, partly - and this is my very personal opinion - the
> problem is that releasing with GCC 4.x has been rushed

I'd have to agree with you on that. I understand the appeal of exciting
press releases but there were over 75 GCC 4.1 bugs still open for
problems in *~arch* when the decision was made to go stable. Even now
there's more than 50 left, with an equal and growing number of stable bugs.

On the other hand, the (misinformed?) perception that Gentoo was
trailing further and further behind the other distros in terms of
version numbers had been raised more than a couple times in the last six
months, so i can see the reasoning behind wanting to make a statement.

--de.


--
gentoo-dev@gentoo.org mailing list