Re: Install of pkg fuse-ntfs fails because of undefined symbol in pkg!?!

[scratch65535]

On Sun, 12 Feb 2017 13:44:10 -0800, "C Hutchinson"
 wrote:

>> I trained as a clinical psychologist, not in computer science or
>> ee.   We learned that adults are notoriously resistant to change
>This is a very astute observation. One I recognized as an absolute;
>some two and a half decades ago, myself. Well done!

Beat y'to it -- my doc training was >4 decades ago.  Neener
neener!  :-)

>
>> (not that we hadn't noticed that ourselves, most of us, but it
>> was nice to have it confirmed "officially").  
>> 
>> But change is possible, even for adults.  The key to change is to
>> realise, at the gut level, that change *is* possible, that things
>> need *not* always be as they are now.   That life really truly
>> can be better. 
>> 
>> That's really hard for most adults to believe.  The prospect of
>> change is terrifying because it threatens us with loss of
>> control.  
>The largest fail in recent history, in this regard was pkg(8).
>Not that pkg itself is bad, per se; but in the way it was
>"presented". Or rather "dropped" like a bomb. Without having
>been "vetted" prior, by those that would now be /required/ to use
>it -- not an option.

That's another side-effect of moving too fast:  the certainty
that sacrificing accountability and buy-in is not only necessary,
but justified.  The stakeholder community becomes a threat to the
success of the program rather than a necessary part of it.  

(It's interesting, also horrible, that the same dynamic operates
in other domains too.  See, e.g., Mayer, M. "They Thought They
Were Free".)


>
>The "magic bullet" is when the "change" is perceived as being a
>change they "wanted", or "asked for".
>FreeBSD has seen *many* requests for change(s) over the years.
>Most of which are dismissed, ignored, or outright rejected.
>Even tho many were variations of the same.
>Solution? /Listen/ to those requests. Perform a pattern match
>to discover similarities in requests, and propose what could
>be perceived "feasible" implementations of those requests. Then
>implement them, and you'll gain a enthusiastic following.
>Period.

Yep.  

The problem with implementing any such solution is that the
majority of technical people --I think I can extrapolate from
those I've met and worked with in my 40+ industry years-- are
focused on tech issues as an end in themselves.  

The idea that the users of their products are mostly not
technical and have very different yet fully legitimate --more
legitimate than their own, really-- desires and needs for the
systems being built is just not very important compared to
solving the hard technical problems they contend with every day.
And so they don't want to know.

 
 
>> But that doesn't mean we're currently doing the right things to
>> regain share from Linux and save FreeBSD.
>
>This is NOT, nor has never has been a "Linux vs BSD" thing. Linux
>is not UNIX. Ahem... Let me rephrase that;
>This is NOT a Linux vs BSD thing, anymore than a Windows vs BSD
>thing.
>This is an "Operating System of choice" thing. If your OS provides
>what a majority of what people need, and want; you're successful,
>and popular. It's as simple as that.

Sure.  But in practice it is Linux vs BSD because those are now
effectively the only 2 Unixoid games in town.

___
freebsd-ports@freebsd.org mailing list
https://lists.freebsd.org/mailman/listinfo/freebsd-ports
To unsubscribe, send any mail to "freebsd-ports-unsubscr...@freebsd.org"

Re: Install of pkg fuse-ntfs fails because of undefined symbol in pkg!?!

[C Hutchinson]

On Sat, 11 Feb 2017 07:31:10 -0500  wrote

> On Fri, 10 Feb 2017 17:46:15 +0100, Kurt Jaeger 
> wrote:
> 
> >Hi!
> 
> Moin! 
> 
> >
> >> On Thu, 9 Feb 2017 17:26:00 +0100, Kurt Jaeger wrote
> >> >Getting the ports/pkg tree moving with the velocity necessary
> >> >to cope with the fast-changing world, sometimes things break
> >> >and we all try to prevent this. Sometimes, mistakes happen...
> >> 
> >> But it's the velocity that's the problem, Kurt.  
> >
> >While I very much sympathize with "The world rotates too fast,
> >I want to get off", for me it looks like as a project we do
> >not have alternatives.
> 
> Why not?  What would happen to fBSD that's not already happening?
> Why aren't people asking what's going on and how to turn it
> around?  Could it be because they're too busy being busy?
> 
> There's a well-known problem that kills ground-attack pilots all
> the time (or it used to; maybe they have safety features built
> into the aircraft now).  They become fixated on their target, and
> they bring the nose of their aircraft further and further down to
> keep the target in their sights.  Which causes them to fly right
> into the ground!
> 
> Fixation is a problem in other fields, too.
> 
> >
> >> Do you know of anyone who has successfully defended, or even
> >> tried to defend, the current manic pace of revision and
> >> obsoleting?
> >
> >Is it defense, if we see many projects (open source etc)
> >shorten their cycle time (e.g. php7), because they see the need to
> >add features or patch security issues (and breaks APIs/ABIs doing either) ?
> 
> It seems more like an excuse than a defence, to me.  Is it
> pushing Linux back?  If not, what *would* push Linux back?  Why
> is Linux so successful even though fBSD is older and better, and
> was once completely dominant in the space?  What are the Linux
> projects doing that we're not?
> 
> >
> >And if we try to keep up and for this, if we add features to the
> >ports framework ? I'm doing this (application mgmt on unix systems)
> >for a long time now, a quarter of a century, and I see no viable
> >alternative in the problem space we work in.
> 
> I trained as a clinical psychologist, not in computer science or
> ee.   We learned that adults are notoriously resistant to change
This is a very astute observation. One I recognized as an absolute;
some two and a half decades ago, myself. Well done!

> (not that we hadn't noticed that ourselves, most of us, but it
> was nice to have it confirmed "officially").  
> 
> But change is possible, even for adults.  The key to change is to
> realise, at the gut level, that change *is* possible, that things
> need *not* always be as they are now.   That life really truly
> can be better. 
> 
> That's really hard for most adults to believe.  The prospect of
> change is terrifying because it threatens us with loss of
> control.  
The largest fail in recent history, in this regard was pkg(8).
Not that pkg itself is bad, per se; but in the way it was
"presented". Or rather "dropped" like a bomb. Without having
been "vetted" prior, by those that would now be /required/ to use
it -- not an option.

The "magic bullet" is when the "change" is perceived as being a
change they "wanted", or "asked for".
FreeBSD has seen *many* requests for change(s) over the years.
Most of which are dismissed, ignored, or outright rejected.
Even tho many were variations of the same.
Solution? /Listen/ to those requests. Perform a pattern match
to discover similarities in requests, and propose what could
be perceived "feasible" implementations of those requests. Then
implement them, and you'll gain a enthusiastic following.
Period.

> 
> But we don't have to lose control.  We can change a bit at a
> time, staying in control all the while.  All we need is the will
> to do it.  (The world is in the mess it's in because, it seems,
> most of those who have the will to change use it for malignant
> purposes --the last several US presidents being cases in point)
> 
> >
> >I also see that this very fast speed uses up huge amounts of
> >person power and compute resources (all those folks rebuilding
> >many ports in their build hosts). But it's not easy to stop off
> >this planet 8-}
> 
> I would never urge that we jump off the planet.  It'd be too hard
> to breathe outside the atmospheric envelope.  :-)
> 
> But that doesn't mean we're currently doing the right things to
> regain share from Linux and save FreeBSD.

This is NOT, nor has never has been a "Linux vs BSD" thing. Linux
is not UNIX. Ahem... Let me rephrase that;
This is NOT a Linux vs BSD thing, anymore than a Windows vs BSD
thing.
This is an "Operating System of choice" thing. If your OS provides
what a majority of what people need, and want; you're successful,
and popular. It's as simple as that.

That's all I have to say in this matter. I generally shy away
from such discussions. As history proves that they devolve into
a "bikeshed(tm)"
 -- 

Re: Install of pkg fuse-ntfs fails because of undefined symbol in pkg!?!

[scratch65535]

[Default] On Sun, 12 Feb 2017 15:30:11 +0100, Michelle Sullivan
 wrote:

>scratch65...@att.net wrote:
>>
>> Developing the original packaging scheme would take some
>> top-notch engineering.
>>
>
>If you're talking the pkg_* tools - already have them working with 
>later/patched software here...  (In fact my ports tree now has some 
>packages more up to date than the FreeBSD one... and with stuff fixed 
>that was just marked BROKEN on FreeBSD 9.x...  Of course thats for my 
>package/patch set so YMMV and there are a lot that are not upgraded to 
>the current FreeBSD set - but then it's only me on it.

ah, sorry Michelle, I didn't quite mean "packaging" as in "pkg
packaging", but more "packaging" as in "integration and polish":
load & go.

Stick the disc in the drive, boot, and a menu appears from which
you pick the functions and flavors you want.  And when it
finishes, you have a running system specialised to the role(s)
you chose.  Better than Microsoft, better than Linux.  

No fears, no tears.  Load and go.
___
freebsd-ports@freebsd.org mailing list
https://lists.freebsd.org/mailman/listinfo/freebsd-ports
To unsubscribe, send any mail to "freebsd-ports-unsubscr...@freebsd.org"

Re: Install of pkg fuse-ntfs fails because of undefined symbol in pkg!?!

[Michelle Sullivan]

Grzegorz Junka wrote:



On Linux companies contribute drivers and dedicated applications and 
it's a win-win situation. More drivers mean people can more easily 
re-use their hardware that they bought for Windows, and more happy 
users means companies are contributing more drivers.


Maybe the fate was set on Linux because FreeBSD was considered as a 
server operating system rather than a Windows replacement, which Linux 
was always trying to be? And good UI/desktop always provides a better 
user experience than a command line and terminal.


And why would people want to switch between FreeBSD on the server and 
Linux/Windows on their desktop if they can just go Linux/Windows all 
the way through? I believe it's only by providing a good desktop 
experience that FreeBSD can survive.


On that we will always be completely opposed in rational... However, I 
knwo where you're coming from but consider this:


Linux has a core team that pretty much work on the kernel and modules 
and have sweet FA to do with the OS/distro.  This is why kernels move 
separately from the distros, unlike FreeBSD.  FreeBSD is a distro so is 
there any wonder why stuff doesn't get fixed and other stuff does but 
only on new versions, and why changing anything in the distro results in 
new versions of everything...


There are cases where this was pretty much a requirement, but only for 
the technical to really know...  Eg when threading support started.. 
when symbol versioning started, in both cases the whole OS had to move 
along with the Kernel... I get that... but what about 7, to 8, to 9, to 
10, to 11, to 12?  pkgng being forced on everyone at 9 to 10 ?  pkgng 
attempting to replace freebsd-update between 10 and 11?  That'd be 
pretty much like RedHat going from 5, to 6, to 7 by going from rpm, to 
rpm+yum, to apt-get+deb... as if that would work!


Perhaps the answer is to separate (as much as possible) the kernel and 
base OS libraries from everything else... Do we really need gcc/clang 
built into the base OS?  Do we need some of the obvious...? (openssl, 
ntp, ssh, inetd, telnet, ftp etc etc etc? - sure some tools need to be 
on the installation media, but really are they needed in the base OS?)


FreeBSD doesn't install a desktop by default, why not separate a Desktop 
team out as Grzegorz seems to think it's of vital importance and would 
come back to competitivness with   linux et al...  make it kernel/distro 
independent ... oh wait it already is!


... ok stopping here, because its turning more into a rant, but I know 
others have seen this point/thought in the same thread... personally I'm 
only interested in the server side of things as I use OSX as my desktop 
which funnily enough is using guess what... a BSD Kernel!.,,


--
Michelle Sullivan
http://www.mhix.org/

___
freebsd-ports@freebsd.org mailing list
https://lists.freebsd.org/mailman/listinfo/freebsd-ports
To unsubscribe, send any mail to "freebsd-ports-unsubscr...@freebsd.org"

Re: Install of pkg fuse-ntfs fails because of undefined symbol in pkg!?!

[Michelle Sullivan]

scratch65...@att.net wrote:


Developing the original packaging scheme would take some
top-notch engineering.



If you're talking the pkg_* tools - already have them working with 
later/patched software here...  (In fact my ports tree now has some 
packages more up to date than the FreeBSD one... and with stuff fixed 
that was just marked BROKEN on FreeBSD 9.x...  Of course thats for my 
package/patch set so YMMV and there are a lot that are not upgraded to 
the current FreeBSD set - but then it's only me on it.


Michelle

--
Michelle Sullivan
http://www.mhix.org/

___
freebsd-ports@freebsd.org mailing list
https://lists.freebsd.org/mailman/listinfo/freebsd-ports
To unsubscribe, send any mail to "freebsd-ports-unsubscr...@freebsd.org"

Re: Install of pkg fuse-ntfs fails because of undefined symbol in pkg!?!

[scratch65535]

[Default] On Sat, 11 Feb 2017 15:11:06 +0100, Kurt Jaeger
 wrote:

>Hi!

Moin! 

>
>> >> But it's the velocity that's the problem, Kurt.  
>
>> >While I very much sympathize with "The world rotates too fast,
>> >I want to get off", for me it looks like as a project we do
>> >not have alternatives.
>> 
>> Why not?  What would happen to fBSD that's not already happening?
>
>Maybe if other folks do not share your analysis that the fbsd world
>is loosing market/mindshare etc, they come to different conclusions
>on what they want from the system.

I'm sure you're right.  But I'd like to see the basis for their
different analysis.  

As I said in response to your other post, I look at most of the
same things I looked at back in the day:  market penetration,
applications, favorable press.  The one thing I don't look at now
are the professional analyses for which companies must splash out
hundreds or thousands of dollars.

So I wonder what they're looking at that would lead them to such
a different conclusion.

>
>> Why aren't people asking what's going on and how to turn it
>> around?  Could it be because they're too busy being busy?
>
>My impression is they are busy innovating, and that is good
>and I thank them for it

But by all the usual metrics (penetration, apps, press) it's not
working.  If we're doing the wrong thing, even doing it Really
Well won't save us.

>
>> >Is it defense, if we see many projects (open source etc)
>> >shorten their cycle time (e.g. php7), because they see the need to
>> >add features or patch security issues (and breaks APIs/ABIs doing either) ?
>> 
>> It seems more like an excuse than a defence, to me.  Is it
>> pushing Linux back?  If not, what *would* push Linux back?
>
>As I asked on the other mail: How do you measure this ?

Penetration, applications, press.  The links that Grzegorz posted
are extremely on-point.  We are in trouble!

>
>[Pathologising others] is not very friendly.

I'm not pathologising anyone.  Fear of change is an ordinary
human response --it's mentioned in scripture, in Shakespeare, and
even  in the US Declaration of Independence!  "[A]ll experience
hath shewn, that mankind are more disposed to suffer, while evils
are sufferable, than to right themselves by abolishing the forms
to which they are accustomed".  Shakespeare:  "[D]read ...
puzzles the will / And makes us rather bear those ills we have /
Than fly to others that we know not of."

But that it's ordinary doesn't mean it's adaptive!  

>
>> But change is possible, even for adults.
>
>So, we're changing a lot and now we hear you complain about those
>changes. 

For some reason, I'm not seeing the "changing a lot".  To me it
looks like the same old Same Old, just madly faster.

>Can you try to describe some changes that would help
>both groups (those that need innovation and those that need
>stability) ?

If I were playing program manager, I'd first map releases to
hardware lifespan:  a major release every five years, and patches
for class A and B bugs or sudden new hardware or package
components (e.g. php 7, mariadb 10) issued ab-necessario.  As
Michelle notes, server hardware can be presumed to be exhausted
after 5 years, so if you've to bring up new boxes anyway, you
might as well upgrade to new bits, too.  But you should not have
to replace your software more often than you replace your
hardware.

If there were enough resources, do a minor release every year.

My second emphasis would be on *predictable* quality, to lure
back businesses (they provided most of our funding and support
back in the day as you know).

Public deliverables would be 3 fully-packaged and
switch-configurable products, 2 of them plug-and-play for people
who just want a reliable tool to use for doing other work, and
the third a high-class kit for people who *don't* just want a
tool to use.  

The tools would be 

- a desktop (as Grzegorz points out, people go with what they
know, and what they know is their desktop.  And right now that
desktop runs Linux),  and

- a server.   

If we couldn't at first manage to package all three, then
priority should be desktop and server before kit.

Cutting back the release frequency and focusing on 3 high-quality
turnkey products should take at least some of the load off the
developers, allowing them to avoid firefighting mode and increase
their free-brain time.

To support quality, I'd mandate, if I were program manager, that
the bits be pulled from their far-flung sources once per cycle
and stored centrally, such that version skew, fnf, and other
errors  would be automatically avoided. 

I'd also mandate that ports be divided into production-quality
bits for the server and desktop packages, and hobby-quality bits.

Between fetching and storing the bits for the build, and focusing
on making sure the bits are indeed production quality for the
packages, there should be many fewer cases of complaints from or
about the software. 

(My install of X complains constantly, even  

Re: Install of pkg fuse-ntfs fails because of undefined symbol in pkg!?!

[Michelle Sullivan]

Mark Linimon wrote:

On Sat, Feb 11, 2017 at 09:51:41AM +0100, Michelle Sullivan wrote:

Tell me  What is the reason for me upgrading those few production
servers from 9.3 to 10/11?... bearing in mind the following:

There isn't any ... oh, except for no new security updates.


But that's the thing isn't it... what security updates? If I take away 
the ports installed I don't actually think there is anything that can 
provide a security issue, at least not one that wouldn't be creating a 
massive ruck around the world... like the ability to hack the FreeBSD 
network stack etc...


The flip side is the more important question.  Why should we ask
developers to support a version of the OS that doesn't have:

 From https://www.freebsd.org/releases/10.0R/relnotes.html:

  * bhyve, drm2, rasbpi, nvme, several overhauled drivers and carp.


..and why can they not appear in 9.x...?



 From https://www.freebsd.org/releases/11.0R/relnotes.html:

  * mpr, mrsas, pms, autofs, other overhauled drivers.


...or these...?

ABI change?  Or is it the same ABI with updated drivers?


I recommend folks reading this thread should examine those links.

As for things removed from base, my understanding is:

  * we have limited people to fix/regression test all the security problems
in base/ software.  A few notorious candidates have been factored out
into ports.


You have less people now I can guarantee you that.



  * there is an overall drive to move to an entirely GPL-free base system.
Perhaps this is not important to you but there are others to whome this
is important.

I have some other thoughts on this but I am being called away to deal with
some other crisis ATM.


The GPL-free base system drive is driving stuff and will change things 
that I can see an OS upgrade/version change is called for... but why not 
get all of it done and put into 10.0 instead of driving forward with 
getting to 12.0 as fast as possible?



Regards,

--
Michelle Sullivan
http://www.mhix.org/

___
freebsd-ports@freebsd.org mailing list
https://lists.freebsd.org/mailman/listinfo/freebsd-ports
To unsubscribe, send any mail to "freebsd-ports-unsubscr...@freebsd.org"

Re: Install of pkg fuse-ntfs fails because of undefined symbol in pkg!?!

[Grzegorz Junka]

On 11/02/2017 15:15, scratch65...@att.net wrote:

On Sat, 11 Feb 2017 15:02:42 +0100, Kurt Jaeger 
wrote:


Hi!


And is the "keeping up with" working, Mark?  Are we regaining
share from Linux?  No.   We're continuing to lose.

How do you measure that ?

Mostly I compare penetration of servers at web-hosting sites,
availability of applications, particularly technical ones, and
favorable mentions in the business and technical press.

In all cases, FreeBSD is an afterthought, if it's mentioned at
all.


https://w3techs.com/technologies/details/os-freebsd/all/all
https://w3techs.com/technologies/details/os-linux/all/all

Also https://news.ycombinator.com/item?id=12199394

Grzegorz
___
freebsd-ports@freebsd.org mailing list
https://lists.freebsd.org/mailman/listinfo/freebsd-ports
To unsubscribe, send any mail to "freebsd-ports-unsubscr...@freebsd.org"

Re: Install of pkg fuse-ntfs fails because of undefined symbol in pkg!?!

[Grzegorz Junka]

On 11/02/2017 12:31, scratch65...@att.net wrote:

On Fri, 10 Feb 2017 17:46:15 +0100, Kurt Jaeger 
wrote:


Hi!

Moin!


On Thu, 9 Feb 2017 17:26:00 +0100, Kurt Jaeger wrote

Getting the ports/pkg tree moving with the velocity necessary
to cope with the fast-changing world, sometimes things break
and we all try to prevent this. Sometimes, mistakes happen...

But it's the velocity that's the problem, Kurt.

While I very much sympathize with "The world rotates too fast,
I want to get off", for me it looks like as a project we do
not have alternatives.

Why not?  What would happen to fBSD that's not already happening?
Why aren't people asking what's going on and how to turn it
around?  Could it be because they're too busy being busy?

There's a well-known problem that kills ground-attack pilots all
the time (or it used to; maybe they have safety features built
into the aircraft now).  They become fixated on their target, and
they bring the nose of their aircraft further and further down to
keep the target in their sights.  Which causes them to fly right
into the ground!

Fixation is a problem in other fields, too.


Do you know of anyone who has successfully defended, or even
tried to defend, the current manic pace of revision and
obsoleting?

Is it defense, if we see many projects (open source etc)
shorten their cycle time (e.g. php7), because they see the need to
add features or patch security issues (and breaks APIs/ABIs doing either) ?

It seems more like an excuse than a defence, to me.  Is it
pushing Linux back?  If not, what *would* push Linux back?  Why
is Linux so successful even though fBSD is older and better, and
was once completely dominant in the space?  What are the Linux
projects doing that we're not?



I think the answer to this question is simple and is the same reason why 
Windows and MacOS have been successful over time - user experience. To 
be more exact - desktop user experience. People like me, developers, use 
computers not only for work but also to do their daily stuff - watch 
DVD, listen to music, catch up with people on Facebook, play games. 
Sometimes I do need to switch to Windows but if I could just use FreeBSD 
I would be done with that. I can do many of these things on FreeBSD now 
but it has always been an uphill path. Everything is much easier on 
Linux. Now we also have PCBSD/TrueOS but it shares the same pain as 
FreeBSD - lack of drivers.


On Linux companies contribute drivers and dedicated applications and 
it's a win-win situation. More drivers mean people can more easily 
re-use their hardware that they bought for Windows, and more happy users 
means companies are contributing more drivers.


Maybe the fate was set on Linux because FreeBSD was considered as a 
server operating system rather than a Windows replacement, which Linux 
was always trying to be? And good UI/desktop always provides a better 
user experience than a command line and terminal.


And why would people want to switch between FreeBSD on the server and 
Linux/Windows on their desktop if they can just go Linux/Windows all the 
way through? I believe it's only by providing a good desktop experience 
that FreeBSD can survive.


Grzegorz
___
freebsd-ports@freebsd.org mailing list
https://lists.freebsd.org/mailman/listinfo/freebsd-ports
To unsubscribe, send any mail to "freebsd-ports-unsubscr...@freebsd.org"

Re: Install of pkg fuse-ntfs fails because of undefined symbol in pkg!?!

[Mark Linimon]

On Sat, Feb 11, 2017 at 09:51:41AM +0100, Michelle Sullivan wrote:
> Tell me  What is the reason for me upgrading those few production
> servers from 9.3 to 10/11?... bearing in mind the following:

There isn't any ... oh, except for no new security updates.

The flip side is the more important question.  Why should we ask
developers to support a version of the OS that doesn't have:

>From https://www.freebsd.org/releases/10.0R/relnotes.html:

 * bhyve, drm2, rasbpi, nvme, several overhauled drivers and carp.

>From https://www.freebsd.org/releases/11.0R/relnotes.html:

 * mpr, mrsas, pms, autofs, other overhauled drivers.

I recommend folks reading this thread should examine those links.

As for things removed from base, my understanding is:

 * we have limited people to fix/regression test all the security problems
   in base/ software.  A few notorious candidates have been factored out
   into ports.

 * there is an overall drive to move to an entirely GPL-free base system.
   Perhaps this is not important to you but there are others to whome this
   is important.

I have some other thoughts on this but I am being called away to deal with
some other crisis ATM.

mcl
___
freebsd-ports@freebsd.org mailing list
https://lists.freebsd.org/mailman/listinfo/freebsd-ports
To unsubscribe, send any mail to "freebsd-ports-unsubscr...@freebsd.org"

Re: Install of pkg fuse-ntfs fails because of undefined symbol in pkg!?!

[Lars Engels]

On Sat, Feb 11, 2017 at 10:15:41AM -0500, scratch65...@att.net wrote:
> On Sat, 11 Feb 2017 15:02:42 +0100, Kurt Jaeger 
> wrote:
> 
> >Hi!
> >
> >> And is the "keeping up with" working, Mark?  Are we regaining
> >> share from Linux?  No.   We're continuing to lose.  
> >
> >How do you measure that ?
> 
> Mostly I compare penetration of servers at web-hosting sites,
> availability of applications, particularly technical ones, and
> favorable mentions in the business and technical press. 
> 
> In all cases, FreeBSD is an afterthought, if it's mentioned at
> all.

Which is mostly because even my grandmother heard of Linux while FreeBSD
is mature and venerable but also mostly unknown even to IT people. It's not
a technical issue in the first place. We just suck at promoting FreeBSD.


pgpJBvhWZzWJc.pgp
Description: PGP signature

Re: Install of pkg fuse-ntfs fails because of undefined symbol in pkg!?!

[Kurt Jaeger]

Hi!

> >> But it's the velocity that's the problem, Kurt.  

> >While I very much sympathize with "The world rotates too fast,
> >I want to get off", for me it looks like as a project we do
> >not have alternatives.
> 
> Why not?  What would happen to fBSD that's not already happening?

Maybe if other folks do not share your analysis that the fbsd world
is loosing market/mindshare etc, they come to different conclusions
on what they want from the system.

> Why aren't people asking what's going on and how to turn it
> around?  Could it be because they're too busy being busy?

My impression is they are busy innovating, and that is good
and I thank them for it.

> >Is it defense, if we see many projects (open source etc)
> >shorten their cycle time (e.g. php7), because they see the need to
> >add features or patch security issues (and breaks APIs/ABIs doing either) ?
> 
> It seems more like an excuse than a defence, to me.  Is it
> pushing Linux back?  If not, what *would* push Linux back?

As I asked on the other mail: How do you measure this ?

[Pathologising others] is not very friendly.

> But change is possible, even for adults.

So, we're changing a lot and now we hear you complain about those
changes. Can you try to describe some changes that would help
both groups (those that need innovation and those that need
stability) ?

> But that doesn't mean we're currently doing the right things to
> regain share from Linux and save FreeBSD.

Is this really a us-vrs-them thing ? I don't see it that way.
There is innovation all across the board and to be able to co-innovation,
certain things need to be done. In the base system and in the
ports tree.

-- 
p...@opsec.eu+49 171 3101372 3 years to go !
___
freebsd-ports@freebsd.org mailing list
https://lists.freebsd.org/mailman/listinfo/freebsd-ports
To unsubscribe, send any mail to "freebsd-ports-unsubscr...@freebsd.org"

Re: Install of pkg fuse-ntfs fails because of undefined symbol in pkg!?!

[Kurt Jaeger]

Hi!

> And is the "keeping up with" working, Mark?  Are we regaining
> share from Linux?  No.   We're continuing to lose.  

How do you measure that ?

-- 
p...@opsec.eu+49 171 3101372 3 years to go !
___
freebsd-ports@freebsd.org mailing list
https://lists.freebsd.org/mailman/listinfo/freebsd-ports
To unsubscribe, send any mail to "freebsd-ports-unsubscr...@freebsd.org"

Re: Install of pkg fuse-ntfs fails because of undefined symbol in pkg!?!

[scratch65535]

On Fri, 10 Feb 2017 17:46:15 +0100, Kurt Jaeger 
wrote:

>Hi!

Moin! 

>
>> On Thu, 9 Feb 2017 17:26:00 +0100, Kurt Jaeger wrote
>> >Getting the ports/pkg tree moving with the velocity necessary
>> >to cope with the fast-changing world, sometimes things break
>> >and we all try to prevent this. Sometimes, mistakes happen...
>> 
>> But it's the velocity that's the problem, Kurt.  
>
>While I very much sympathize with "The world rotates too fast,
>I want to get off", for me it looks like as a project we do
>not have alternatives.

Why not?  What would happen to fBSD that's not already happening?
Why aren't people asking what's going on and how to turn it
around?  Could it be because they're too busy being busy?

There's a well-known problem that kills ground-attack pilots all
the time (or it used to; maybe they have safety features built
into the aircraft now).  They become fixated on their target, and
they bring the nose of their aircraft further and further down to
keep the target in their sights.  Which causes them to fly right
into the ground!

Fixation is a problem in other fields, too.

>
>> Do you know of anyone who has successfully defended, or even
>> tried to defend, the current manic pace of revision and
>> obsoleting?
>
>Is it defense, if we see many projects (open source etc)
>shorten their cycle time (e.g. php7), because they see the need to
>add features or patch security issues (and breaks APIs/ABIs doing either) ?

It seems more like an excuse than a defence, to me.  Is it
pushing Linux back?  If not, what *would* push Linux back?  Why
is Linux so successful even though fBSD is older and better, and
was once completely dominant in the space?  What are the Linux
projects doing that we're not?

>
>And if we try to keep up and for this, if we add features to the
>ports framework ? I'm doing this (application mgmt on unix systems)
>for a long time now, a quarter of a century, and I see no viable
>alternative in the problem space we work in.

I trained as a clinical psychologist, not in computer science or
ee.   We learned that adults are notoriously resistant to change
(not that we hadn't noticed that ourselves, most of us, but it
was nice to have it confirmed "officially").  

But change is possible, even for adults.  The key to change is to
realise, at the gut level, that change *is* possible, that things
need *not* always be as they are now.   That life really truly
can be better. 

That's really hard for most adults to believe.  The prospect of
change is terrifying because it threatens us with loss of
control.  

But we don't have to lose control.  We can change a bit at a
time, staying in control all the while.  All we need is the will
to do it.  (The world is in the mess it's in because, it seems,
most of those who have the will to change use it for malignant
purposes --the last several US presidents being cases in point)

>
>I also see that this very fast speed uses up huge amounts of
>person power and compute resources (all those folks rebuilding
>many ports in their build hosts). But it's not easy to stop off
>this planet 8-}

I would never urge that we jump off the planet.  It'd be too hard
to breathe outside the atmospheric envelope.  :-)

But that doesn't mean we're currently doing the right things to
regain share from Linux and save FreeBSD.
___
freebsd-ports@freebsd.org mailing list
https://lists.freebsd.org/mailman/listinfo/freebsd-ports
To unsubscribe, send any mail to "freebsd-ports-unsubscr...@freebsd.org"

Re: Install of pkg fuse-ntfs fails because of undefined symbol in pkg!?!

[Michelle Sullivan]

Mark Linimon wrote:

On Fri, Feb 10, 2017 at 11:02:41PM +0100, Michelle Sullivan wrote:

Or the last strong hold you have - the server owners - get so p**sed off in
reality they can't keep up with the OS updates that they migrate away...

So we should give up on EFI, 4k drives, and SSD?


Not what I said, suggested or thought.  There is a difference between 
adding drivers and installation/boot tools and an incessant push to get 
people to upgrade every (what seems like 5 minutes)... yes I know it's 
longer than that, it might be a year, it might be two, but the hardware 
is outlasting (by a long way) the OS.


Tell me  What is the reason for me upgrading those few production 
servers from 9.3 to 10/11?... baring in mind the following:


The tools/utilities in the base OS (ntp, ssh, openssl, etc etc etc) have 
been removed/disabled.
The ports I am using to replace the base utilities is still using pkg_* 
tools and is up to date for the 1000 odd packages I have across my 
environment.
The environment does not contain any new hardware that is not already 
supported.


Then after that consider this:

On some of my servers I have been attempting to move services to using 
'powerdns' which was marked as broken on 9.3 since August 2016...  It 
failed to link with 'to_string' not a part of 'std' ... however with a 
little research and some help (as I know very little about c++) it turns 
out that the problem that caused it is the same problem that has been 
present since 2014 on 8.x and was reported on the mailing lists along 
with a fix... oh and since then it's been reported on 10.x as well... 
and still no fix...


... The fix it turns out was very simple it's a bad #ifdef/missing 
#define in the C++ headers/config ...


Guess what... it's fixed on my systems, but get "9.3 is obsolete, please 
upgrade" if I ask for a fix or even suggest about patching the fix in


Its not the only thing, but it shows the sentiment, if you're not 
running HEAD, then *most* are not interested in helping you...  This is 
the reason I forked...  This is the reason why linux has so many 
variants, this is why FreeBSD is either now, or heading to be "just 
another distro" ... we're better off with DeadRat, we can pay support 
put it on our servers and know that it'll be supported *and patched* for 
the 5+ years we'll have the hardware... and this is what I was fighting 
within the company but they're right FreeBSD is not a viable option for 
the company as the OS moves too fast for either our DCs or the on prem 
embedded hardware we're a security company, we can't afford to be 
running an unsupported OS, and we can't afford to have 90% of the 
company staff admins so that we can keep upgrading our DCs and all the 
appliances.


I respect you Mark, have seen you about here for years now, you're not 
stupid, but you have to see where the forcing of upgrades and drive to 
keep up with the joneses is hurting the userbase and therefore FreeBSD 
as a whole and just as a parting shot, about to attempt to deploy my 
fork onto some Sun hardware and if it works (which I expect it to) I'm 
going to try the same fix on there to see if that also fixes the 
build/link problem.


--
Michelle Sullivan
http://www.mhix.org/

___
freebsd-ports@freebsd.org mailing list
https://lists.freebsd.org/mailman/listinfo/freebsd-ports
To unsubscribe, send any mail to "freebsd-ports-unsubscr...@freebsd.org"

Re: Install of pkg fuse-ntfs fails because of undefined symbol in pkg!?!

[Mark Linimon]

On Fri, Feb 10, 2017 at 11:02:41PM +0100, Michelle Sullivan wrote:
> Or the last strong hold you have - the server owners - get so p**sed off in
> reality they can't keep up with the OS updates that they migrate away...

So we should give up on EFI, 4k drives, and SSD?

mcl
___
freebsd-ports@freebsd.org mailing list
https://lists.freebsd.org/mailman/listinfo/freebsd-ports
To unsubscribe, send any mail to "freebsd-ports-unsubscr...@freebsd.org"

Re: Install of pkg fuse-ntfs fails because of undefined symbol in pkg!?!

[Michelle Sullivan]

Mark Linimon wrote:

On Fri, Feb 10, 2017 at 08:17:31AM -0500, scratch65...@att.net wrote:

A good rule of thumb from industry in the case of major software
would be "forever", meaning until it's very unlikely that anyone
is still using it because of hardware obsolescence, etc.

(Sigh.)  And how many people do you think it takes to do such support?


Why is Linux able to so easily replace FreeBSD?  The desktop is
gone.  Servers are going.  The new AMD chips are being tested
against Intel on Linux boxes, not FreeBSD boxes.  FreeBSD is
being made obsolete.

In other words, if we move fast enough to try to keep up with Linux
changes, FreeBSD is obsolete.  If we move more slowly than Linux, then
FreeBSD is obsolete.

I'm being serious.  We get criticized either way.

Also, for package sets, consider that size * each OS release * each
architecture (ok, some architectures) = a lot of disk space.  We
simply have finite disk space.

IMHO, the days that we can expect ports maintainers and committers to
keep e.g. a FreeBSD 4.11 viable for years are over.  By the EOL of 4.11,
we were asking volunteers to support *4* major OS releases.  That was
crazy.

As for the OS releases, we're trying to keep up with new disk technologies,
new ways of booting, new wireless techniques, graphics APIs that change
rapidly, and on and on.  The pace of these changes is outside our control.
We can keep up or become irrelevant.

Or the last strong hold you have - the server owners - get so p**sed off 
in reality they can't keep up with the OS updates that they migrate away...


FreeBSD seems to have entered (trying to enter) the desktop market... 
good luck competing with Apple and Microsoft... I guess FreeBSD will be 
the next Solaris.


--
Michelle Sullivan
http://www.mhix.org/

___
freebsd-ports@freebsd.org mailing list
https://lists.freebsd.org/mailman/listinfo/freebsd-ports
To unsubscribe, send any mail to "freebsd-ports-unsubscr...@freebsd.org"

Re: Install of pkg fuse-ntfs fails because of undefined symbol in pkg!?!

[Mark Linimon]

On Fri, Feb 10, 2017 at 08:17:31AM -0500, scratch65...@att.net wrote:
> A good rule of thumb from industry in the case of major software
> would be "forever", meaning until it's very unlikely that anyone
> is still using it because of hardware obsolescence, etc.

(Sigh.)  And how many people do you think it takes to do such support?

> Why is Linux able to so easily replace FreeBSD?  The desktop is
> gone.  Servers are going.  The new AMD chips are being tested
> against Intel on Linux boxes, not FreeBSD boxes.  FreeBSD is
> being made obsolete.

In other words, if we move fast enough to try to keep up with Linux
changes, FreeBSD is obsolete.  If we move more slowly than Linux, then
FreeBSD is obsolete.

I'm being serious.  We get criticized either way.

Also, for package sets, consider that size * each OS release * each
architecture (ok, some architectures) = a lot of disk space.  We
simply have finite disk space.

IMHO, the days that we can expect ports maintainers and committers to
keep e.g. a FreeBSD 4.11 viable for years are over.  By the EOL of 4.11,
we were asking volunteers to support *4* major OS releases.  That was
crazy.

As for the OS releases, we're trying to keep up with new disk technologies,
new ways of booting, new wireless techniques, graphics APIs that change
rapidly, and on and on.  The pace of these changes is outside our control.
We can keep up or become irrelevant.

mcl
___
freebsd-ports@freebsd.org mailing list
https://lists.freebsd.org/mailman/listinfo/freebsd-ports
To unsubscribe, send any mail to "freebsd-ports-unsubscr...@freebsd.org"

Re: Install of pkg fuse-ntfs fails because of undefined symbol in pkg!?!

[Kurt Jaeger]

Hi!

> On Thu, 9 Feb 2017 17:26:00 +0100, Kurt Jaeger wrote
> >Getting the ports/pkg tree moving with the velocity necessary
> >to cope with the fast-changing world, sometimes things break
> >and we all try to prevent this. Sometimes, mistakes happen...
> 
> But it's the velocity that's the problem, Kurt.  

While I very much sympathize with "The world rotates too fast,
I want to get off", for me it looks like as a project we do
not have alternatives.

> Do you know of anyone who has successfully defended, or even
> tried to defend, the current manic pace of revision and
> obsoleting?

Is it defense, if we see many projects (open source etc)
shorten their cycle time (e.g. php7), because they see the need to
add features or patch security issues (and breaks APIs/ABIs doing either) ?

And if we try to keep up and for this, if we add features to the
ports framework ? I'm doing this (application mgmt on unix systems)
for a long time now, a quarter of a century, and I see no viable
alternative in the problem space we work in.

I also see that this very fast speed uses up huge amounts of
person power and compute resources (all those folks rebuilding
many ports in their build hosts). But it's not easy to stop off
this planet 8-}

-- 
p...@opsec.eu+49 171 3101372 3 years to go !
___
freebsd-ports@freebsd.org mailing list
https://lists.freebsd.org/mailman/listinfo/freebsd-ports
To unsubscribe, send any mail to "freebsd-ports-unsubscr...@freebsd.org"

Re: Install of pkg fuse-ntfs fails because of undefined symbol in pkg!?!

[scratch65535]

On Thu, 9 Feb 2017 10:09:35 -0500, Steve Wills
 wrote:

>Hi,
>
>On 02/08/2017 12:34, scratch65...@att.net wrote:
>> 
>> I *did* check for bug reports.  I did a search on "utimenstat"
>> and found exactly one, which had been withdrawn as not being a
>> bug. 
>> 
>> But it *is* a bug.  It's a bug on several levels, the most
>> significant of which is that the overly frantic schedule makes
>> versions have the lifespan of a mayfly.   And we're told "just
>> upgrade", as though there's some physical law mandating the
>> craziness.
>
>Ports and packages are maintained on the assumption that the user is
>using a supported version of the OS. We didn't decide when to end
>support for 10.1 or 10.2. How long after the end of life for 10.1 would
>you have ports maintain support?

A good rule of thumb from industry in the case of major software
would be "forever", meaning until it's very unlikely that anyone
is still using it because of hardware obsolescence, etc.
"Support" for out-of-rev software need be no more than disc
space, if that's all that can be afforded.   I'd have been happy
to get that package I thought I would get--it never occurred to
me that the packages for 10.2 would all be summarily purged!



>
>> There are people for whom the system is a tool, not a hobby. They
>> don't want to have to rebuild their tools any more than
>> carpenters want to replace their hammers and levels every year or
>> two.  
>
>If you've having trouble upgrading that are causing you to rebuild, then
>that's a different issue, but not one I can help with. It doesn't change
>the fact that we don't support unsupported versions of the OS.

But the transition to "unsupported" is not a function of physical
law.  It's a human choice, and can be revisited any time people
are willing to do so.

Why is Linux able to so easily replace FreeBSD?  The desktop is
gone.  Servers are going.  The new AMD chips are being tested
against Intel on Linux boxes, not FreeBSD boxes.  FreeBSD is
being made obsolete.

This  not happening by accident.   And it won't stop by itself.  


>
>> For those people (I'm one) long version lifespans and bug-free
>> operation  is a much bigger desideratum than winning the secret
>> race (I presume there is some kind of secret race going on, since
>> otherwise the crazy scheduling makes No Sense At All).  I can't
>> work out what the strategy for winning is, if there is a
>> strategy, but I do know that it's not working.   Linux has all
>> but won already, and that's sickening.
>
>Ports are maintained by volunteers. If you would like to volunteer to
>support branches for longer periods of time, let's talk about that.

I'm a customer.   I'm one of the people for whom the dev and
porting work is supposedly being done.

The problem is the lunatic turnover rate.  It prevents things
from being done in any kind of measured way.  It's all
firefighting, all the time.   That guarantees that things fall
through the cracks, tempers get short, and people burn out.  Who
benefits?


>
>> I've been using the o/s since before v2 (I still have the cds)
>> and have watched FreeBSD go from being the leading Unix on Intel
>> boxes to all-but-dead.  I don't know how to express how saddened
>> I feel about that.
>
>I think ports are really improving and the rate of improvement is going up.

Then why is Linux everywhere and fBSD is circling the drain?
___
freebsd-ports@freebsd.org mailing list
https://lists.freebsd.org/mailman/listinfo/freebsd-ports
To unsubscribe, send any mail to "freebsd-ports-unsubscr...@freebsd.org"

Re: Install of pkg fuse-ntfs fails because of undefined symbol in pkg!?!

[scratch65535]

On Thu, 9 Feb 2017 17:26:00 +0100, Kurt Jaeger wrote

>Getting the ports/pkg tree moving with the velocity necessary
>to cope with the fast-changing world, sometimes things break
>and we all try to prevent this. Sometimes, mistakes happen...

But it's the velocity that's the problem, Kurt.  

Do you know of anyone who has successfully defended, or even
tried to defend, the current manic pace of revision and
obsoleting?  I haven't, but I'd like to read it if you know of an
attempted defence!
___
freebsd-ports@freebsd.org mailing list
https://lists.freebsd.org/mailman/listinfo/freebsd-ports
To unsubscribe, send any mail to "freebsd-ports-unsubscr...@freebsd.org"

Re: Install of pkg fuse-ntfs fails because of undefined symbol in pkg!?!

[Matthew Seaman]

On 02/09/17 16:44, Miroslav Lachman wrote:
> Why don't add some check in to "pkg" to deny (or warn user) upgrade or
> install on unsupported / EOLed system?
> Just check version on current system against some metadata info in
> repository.

Actually the metadata should be in the package, rather than the
repository.  We need to record the OS version the package was compiled
under at the point the package is created, and then pkg(8) can compare
that to the OS version at install time.  This will work not just for the
FreeBSD pkg repos, but for packages built for private repos too.  And it
will still work, even if you grab a bunch of packages from somewhere
else and make your own repo from them.

Even so, pkg(8) should not refuse to install the newer package on the
older system; just emit a big fat warning that what you're doing is
dangerous, and may lead you into regret and grief.  (Unix has a
tradition of not stopping users from doing stupid things, because that
also makes it possible to do amazingly clever things...)

This is complicated by such things as 'NO_ARCH' packages -- your pure
perl/ruby/python code is still going to work almost regardless of the OS
version.  As will all sorts of type fonts or collections of desktop
icons and so forth.

Plus this will need to be carefully debugged when packages are
cross-compiled or compiled in jails on a host system with a different
kernel version.

Cheers,

Matthew




signature.asc
Description: OpenPGP digital signature

Re: Install of pkg fuse-ntfs fails because of undefined symbol in pkg!?!

[Franco Fichtner]

> On 9 Feb 2017, at 6:03 PM, Steve Wills  wrote:
> 
> Just because you don't use any features of the newer version doesn't
> mean it's safe to run binaries built for the newer version on the older
> version, as far as I understand it.

True.  :)

Yet the reports are for missing symbols in pkg and how to fix.  It
beats pulling a ports tree and rebuilding pkg which may or may not
end up being replaced by a newer repo version on the next pkg upgrade.

Plus you still get to garbage-collect compatibility features during
major upgrade jumps.

Fixing this issue in 25k at the same times ends up delaying a workable
solution, whatever it may be.

pkg first, then the rest, no?


Cheers,
Franco
___
freebsd-ports@freebsd.org mailing list
https://lists.freebsd.org/mailman/listinfo/freebsd-ports
To unsubscribe, send any mail to "freebsd-ports-unsubscr...@freebsd.org"

Re: Install of pkg fuse-ntfs fails because of undefined symbol in pkg!?!

[Konstantin Belousov]

On Thu, Feb 09, 2017 at 04:42:45PM +, Matthew Seaman wrote:
> Why do you think it is not being enforced?  Forwards compatibility means
> that during the lifetime of a major branch you can only *add* symbols to
> the system shared libraries, not remove them nor modify any existing
> symbols.  The project has held to that for many years -- not breaking
> ABI forwards compatibility is a really big deal amongst developers.

We try hard to not break ABI backward compatibility between branches
as well, at least for user code (see below). In particular, versioned
libraries in base must be fully backward compatible between branches.
Whole set of the base C runtime libraries is versioned, i.e.
rtld/libc/libthr/libm. libc++ is versioned as well.

For non-versioned libraries, our promise is that on ABI breakage of
any kind, the library version (the libXXX.so.N, the N part) is bumped
so backward compatibility can be provided in some form by installing
previous version of the library. This is done, besides other means, by
misc/compatX ports.

The explanation above is of course, simplified, and somewhat incorrect.
To make it correct would require amount of work which is apparently too
much for single person to do and still be sane.  You can see it that
the project' ABI promise is not formalized even on wiki.

Place were ABI is very badly broken regularly is the management interfaces.
For instance, you are almost guaranteed that ifconfig(8) from a major
branch works only with the kernels on the same branch, and even then,
you must have the binary built with headers from very close kernel
version.  Same for cam(4).

Formalizing these exceptions is the hard part of writing the
ABI guarantee document.
___
freebsd-ports@freebsd.org mailing list
https://lists.freebsd.org/mailman/listinfo/freebsd-ports
To unsubscribe, send any mail to "freebsd-ports-unsubscr...@freebsd.org"

Re: Install of pkg fuse-ntfs fails because of undefined symbol in pkg!?!

[Steve Wills]

Hi,

On 02/09/2017 12:00, Franco Fichtner wrote:
> 
> Let me try another way:
> 
> Since pkg has feature macros for building correctly on different
> FreeBSD versions, namely 10.0, 10.1, 10.2 and 10.3, the way to
> provide binaries without missing symbols is to build pkg with a
> fixed set of feature macros for 10.0.
> 
> I've done this for projects to retain upgrade paths.  It's not
> hard.  It doesn't violate a policy or promise FreeBSD makes, does
> it?
> 

Just because you don't use any features of the newer version doesn't
mean it's safe to run binaries built for the newer version on the older
version, as far as I understand it.

Steve




signature.asc
Description: OpenPGP digital signature

Re: Install of pkg fuse-ntfs fails because of undefined symbol in pkg!?!

[Steve Wills]

Hi,

On 02/09/2017 11:44, Miroslav Lachman wrote:
> Why don't add some check in to "pkg" to deny (or warn user) upgrade or
> install on unsupported / EOLed system?
> Just check version on current system against some metadata info in
> repository.

I would be happy to see a patch that showed how this might be done.

> Is it too much to ask?

It's always too much to ask another to do work for you for free. :)

Steve



signature.asc
Description: OpenPGP digital signature

Re: Install of pkg fuse-ntfs fails because of undefined symbol in pkg!?!

[Franco Fichtner]

> On 9 Feb 2017, at 5:53 PM, Steve Wills  wrote:
> 
> What would enforcement look like? Something like "Sorry, you can't pkg
> update because this system isn't supported any more."? But how would
> that be possible without also breaking things for those who build/ship
> their own OS and packages?

Let me try another way:

Since pkg has feature macros for building correctly on different
FreeBSD versions, namely 10.0, 10.1, 10.2 and 10.3, the way to
provide binaries without missing symbols is to build pkg with a
fixed set of feature macros for 10.0.

I've done this for projects to retain upgrade paths.  It's not
hard.  It doesn't violate a policy or promise FreeBSD makes, does
it?


Cheers,
Franco
___
freebsd-ports@freebsd.org mailing list
https://lists.freebsd.org/mailman/listinfo/freebsd-ports
To unsubscribe, send any mail to "freebsd-ports-unsubscr...@freebsd.org"

Re: Install of pkg fuse-ntfs fails because of undefined symbol in pkg!?!

[Franco Fichtner]

> On 9 Feb 2017, at 5:53 PM, Steve Wills  wrote:
> 
> What would enforcement look like? Something like "Sorry, you can't pkg
> update because this system isn't supported any more."? But how would
> that be possible without also breaking things for those who build/ship
> their own OS and packages?

Let me try another way:

Since pkg has feature macros for building correctly on different
FreeBSD versions, namely 10.0, 10.1, 10.2 and 10.3, the way to
provide binaries without missing symbols is to build pkg with a
fixed set of feature macros for 10.0.

I've done this for projects to retain upgrade paths.  It's not
hard.  It doesn't violate a policy or promise FreeBSD makes, does
it?


Cheers,
Franco
___
freebsd-ports@freebsd.org mailing list
https://lists.freebsd.org/mailman/listinfo/freebsd-ports
To unsubscribe, send any mail to "freebsd-ports-unsubscr...@freebsd.org"

Re: Install of pkg fuse-ntfs fails because of undefined symbol in pkg!?!

[Konstantin Belousov]

On Thu, Feb 09, 2017 at 05:26:00PM +0100, Kurt Jaeger wrote:
> Hi!
> 
> > On Thu, Feb 09, 2017 at 04:30:20PM +0100, Franco Fichtner wrote:
> > > FreeBSD package management makes an ABI promise in the form of
> > > "FreeBSD:10:amd64", but not even pkg code itself adheres to this,
> > > and thus we have had subtle and yet fatal breakage in 10.2 and 10.3.
> > Stop spreading FUD. There is no ABI breakage on stable/10 branch,
> > nor there is a breakage in the package sets.
> 
> On the one hand:
> 
> Maybe we can agree that the pkg binary breaking between different
> 10.x versions was unfortunate ? I understand that it was not a
> break of the ABI promises per se, but I can tell you, I was surprised
> as well, when it bite me 8-}
The pkg did not break.  Your usage of it and probably assumptions which
lead to that usage, are broken.  It is as simple as never use a binary
which was built for later system, on earlier.  Even if the binary run, you
get undefined results, e.g. data corruption.

We even grow kern.disallow_high_osrel knob to help people, who cannot
manage herself, to avoid shooting into their foot.  There are some obvious
drawbacks from setting this knob on by default, which explains why it
is not set (it makes recovering from other user mistakes much harder
and the failure mode much more fatal).

> 
> On the other hand:
> 
> Getting the ports/pkg tree moving with the velocity necessary
> to cope with the fast-changing world, sometimes things break
> and we all try to prevent this. Sometimes, mistakes happen...
> 
> -- 
> p...@opsec.eu+49 171 3101372 3 years to 
> go !
___
freebsd-ports@freebsd.org mailing list
https://lists.freebsd.org/mailman/listinfo/freebsd-ports
To unsubscribe, send any mail to "freebsd-ports-unsubscr...@freebsd.org"

Re: Install of pkg fuse-ntfs fails because of undefined symbol in pkg!?!

[Steve Wills]

Hi,

On 02/09/2017 11:24, Franco Fichtner wrote:
> 
>> On 9 Feb 2017, at 5:21 PM, Steve Wills  wrote:
>>
>> We provide backwards compatibility, not forwards compatibility.
> 
> But don't you see that users won't know this?

Users who don't know their software is no longer supported and refuse to
update can't be helped, IMHO.

> This is a good theory, yet it is difficult in practice because it is
> not being enforced.

What would enforcement look like? Something like "Sorry, you can't pkg
update because this system isn't supported any more."? But how would
that be possible without also breaking things for those who build/ship
their own OS and packages?

Steve




signature.asc
Description: OpenPGP digital signature

Re: Install of pkg fuse-ntfs fails because of undefined symbol in pkg!?!

[Miroslav Lachman]

Kurt Jaeger wrote on 2017/02/09 17:26:

Hi!


On Thu, Feb 09, 2017 at 04:30:20PM +0100, Franco Fichtner wrote:

FreeBSD package management makes an ABI promise in the form of
"FreeBSD:10:amd64", but not even pkg code itself adheres to this,
and thus we have had subtle and yet fatal breakage in 10.2 and 10.3.

Stop spreading FUD. There is no ABI breakage on stable/10 branch,
nor there is a breakage in the package sets.


On the one hand:

Maybe we can agree that the pkg binary breaking between different
10.x versions was unfortunate ? I understand that it was not a
break of the ABI promises per se, but I can tell you, I was surprised
as well, when it bite me 8-}

On the other hand:

Getting the ports/pkg tree moving with the velocity necessary
to cope with the fast-changing world, sometimes things break
and we all try to prevent this. Sometimes, mistakes happen...


I don't have a problem with pkg / ports but I see the point from some 
others perspective.


Why don't add some check in to "pkg" to deny (or warn user) upgrade or 
install on unsupported / EOLed system?
Just check version on current system against some metadata info in 
repository.


Is it too much to ask?

Miroslav Lachman

___
freebsd-ports@freebsd.org mailing list
https://lists.freebsd.org/mailman/listinfo/freebsd-ports
To unsubscribe, send any mail to "freebsd-ports-unsubscr...@freebsd.org"

Re: Install of pkg fuse-ntfs fails because of undefined symbol in pkg!?!

[Matthew Seaman]

On 2017/02/09 16:24, Franco Fichtner wrote:
>> On 9 Feb 2017, at 5:21 PM, Steve Wills  wrote:
>>
>> We provide backwards compatibility, not forwards compatibility.

> But don't you see that users won't know this?

Forward compatibility has been the ABI stability guarantee basically
ever since there has been a FreeBSD project.   Anything you compile now
will continue to work on later OS versions.  However you cannot
guarantee it will work on earlier versions.

The fact that newer binaries frequently will still work on older
releases in no way invalidates this claim.  That's just a reflection of
the fact that the system ABIs do not change particularly frequently.

> This is a good theory, yet it is difficult in practice because it is
> not being enforced.

Why do you think it is not being enforced?  Forwards compatibility means
that during the lifetime of a major branch you can only *add* symbols to
the system shared libraries, not remove them nor modify any existing
symbols.  The project has held to that for many years -- not breaking
ABI forwards compatibility is a really big deal amongst developers.

Cheers,

Matthew




signature.asc
Description: OpenPGP digital signature

Re: Install of pkg fuse-ntfs fails because of undefined symbol in pkg!?!

[Kurt Jaeger]

Hi!

> On Thu, Feb 09, 2017 at 04:30:20PM +0100, Franco Fichtner wrote:
> > FreeBSD package management makes an ABI promise in the form of
> > "FreeBSD:10:amd64", but not even pkg code itself adheres to this,
> > and thus we have had subtle and yet fatal breakage in 10.2 and 10.3.
> Stop spreading FUD. There is no ABI breakage on stable/10 branch,
> nor there is a breakage in the package sets.

On the one hand:

Maybe we can agree that the pkg binary breaking between different
10.x versions was unfortunate ? I understand that it was not a
break of the ABI promises per se, but I can tell you, I was surprised
as well, when it bite me 8-}

On the other hand:

Getting the ports/pkg tree moving with the velocity necessary
to cope with the fast-changing world, sometimes things break
and we all try to prevent this. Sometimes, mistakes happen...

-- 
p...@opsec.eu+49 171 3101372 3 years to go !
___
freebsd-ports@freebsd.org mailing list
https://lists.freebsd.org/mailman/listinfo/freebsd-ports
To unsubscribe, send any mail to "freebsd-ports-unsubscr...@freebsd.org"

Re: Install of pkg fuse-ntfs fails because of undefined symbol in pkg!?!

[Franco Fichtner]

> On 9 Feb 2017, at 5:21 PM, Steve Wills  wrote:
> 
> We provide backwards compatibility, not forwards compatibility.

But don't you see that users won't know this?

This is a good theory, yet it is difficult in practice because it is
not being enforced.


Cheers,
Franco
___
freebsd-ports@freebsd.org mailing list
https://lists.freebsd.org/mailman/listinfo/freebsd-ports
To unsubscribe, send any mail to "freebsd-ports-unsubscr...@freebsd.org"

Re: Install of pkg fuse-ntfs fails because of undefined symbol in pkg!?!

[Franco Fichtner]

> On 9 Feb 2017, at 5:21 PM, Steve Wills  wrote:
> 
> We provide backwards compatibility, not forwards compatibility.

But don't you see that users won't know this?

This is a good theory, yet it is difficult in practice because it is
not being enforced.


Cheers,
Franco
___
freebsd-ports@freebsd.org mailing list
https://lists.freebsd.org/mailman/listinfo/freebsd-ports
To unsubscribe, send any mail to "freebsd-ports-unsubscr...@freebsd.org"

Re: Install of pkg fuse-ntfs fails because of undefined symbol in pkg!?!

[Steve Wills]

Hi,

On 02/09/2017 11:14, Franco Fichtner wrote:
> 
> You're contradicting yourself here.  Either it's compatible or it isn't?
> 

Not at all. There's a difference between backwards compatibility (binary
built on older release works on newer release) and forwards
compatibility (binary built on newer release works on older release).

We provide backwards compatibility, not forwards compatibility.

Steve



signature.asc
Description: OpenPGP digital signature

Re: Install of pkg fuse-ntfs fails because of undefined symbol in pkg!?!

[Franco Fichtner]

> On 9 Feb 2017, at 5:12 PM, Konstantin Belousov  wrote:
> 
> On Thu, Feb 09, 2017 at 04:30:20PM +0100, Franco Fichtner wrote:
>> FreeBSD package management makes an ABI promise in the form of
>> "FreeBSD:10:amd64", but not even pkg code itself adheres to this,
>> and thus we have had subtle and yet fatal breakage in 10.2 and 10.3.
> Stop spreading FUD. There is no ABI breakage on stable/10 branch,
> nor there is a breakage in the package sets. We only promise
> backward-compatibility, and this works as advertized. A binary compiled
> on later system, is not guaranteed to work on the early system even on
> the same branch.
> 
> The current package set for stable/10 is built on 10.3 and is only
> guaranteed to work on 10.3 and later. Trying to make arbitrary
> combinations of binaries and base systems outside of the scope of the
> project.

You're contradicting yourself here.  Either it's compatible or it isn't?


Cheers,
Franco
___
freebsd-ports@freebsd.org mailing list
https://lists.freebsd.org/mailman/listinfo/freebsd-ports
To unsubscribe, send any mail to "freebsd-ports-unsubscr...@freebsd.org"

Re: Install of pkg fuse-ntfs fails because of undefined symbol in pkg!?!

[Konstantin Belousov]

On Thu, Feb 09, 2017 at 04:30:20PM +0100, Franco Fichtner wrote:
> FreeBSD package management makes an ABI promise in the form of
> "FreeBSD:10:amd64", but not even pkg code itself adheres to this,
> and thus we have had subtle and yet fatal breakage in 10.2 and 10.3.
Stop spreading FUD. There is no ABI breakage on stable/10 branch,
nor there is a breakage in the package sets. We only promise
backward-compatibility, and this works as advertized. A binary compiled
on later system, is not guaranteed to work on the early system even on
the same branch.

The current package set for stable/10 is built on 10.3 and is only
guaranteed to work on 10.3 and later. Trying to make arbitrary
combinations of binaries and base systems outside of the scope of the
project.
___
freebsd-ports@freebsd.org mailing list
https://lists.freebsd.org/mailman/listinfo/freebsd-ports
To unsubscribe, send any mail to "freebsd-ports-unsubscr...@freebsd.org"

Re: Install of pkg fuse-ntfs fails because of undefined symbol in pkg!?!

[Steve Wills]

Hi,

On 02/09/2017 11:01, Franco Fichtner wrote:
> 
>> On 9 Feb 2017, at 4:47 PM, Steve Wills  wrote:
>>
>> They're supposed to upgrade to a supported version of FreeBSD.
> 
> pkg won't refuse the upgrade.  And at least if it upgraded, it
> should not break itself.

Even if the update of pkg were done before the upgrade of the OS, if the
user ran "freebsd-update" to upgrade to 10.3, the version of pkg that it
updated to would work properly.

> Imagine a GUI-driven appliance being bricked.  There is nobody
> who can tell it "fetch ports and build pkg to keep using it".

Vendors should be building their own packages.

> Don't get me wrong.  Automation around pkg is really good, though
> that doesn't warrant it's perfect (yet).

I agree it's good, and nothing is ever perfect.

Steve



signature.asc
Description: OpenPGP digital signature

Re: Install of pkg fuse-ntfs fails because of undefined symbol in pkg!?!

[Franco Fichtner]

> On 9 Feb 2017, at 4:47 PM, Steve Wills  wrote:
> 
> They're supposed to upgrade to a supported version of FreeBSD.

pkg won't refuse the upgrade.  And at least if it upgraded, it
should not break itself.

Imagine a GUI-driven appliance being bricked.  There is nobody
who can tell it "fetch ports and build pkg to keep using it".

Don't get me wrong.  Automation around pkg is really good, though
that doesn't warrant it's perfect (yet).


Cheers,
Franco
___
freebsd-ports@freebsd.org mailing list
https://lists.freebsd.org/mailman/listinfo/freebsd-ports
To unsubscribe, send any mail to "freebsd-ports-unsubscr...@freebsd.org"

Re: Install of pkg fuse-ntfs fails because of undefined symbol in pkg!?!

[Steve Wills]

Hi,

On 02/09/2017 10:30, Franco Fichtner wrote:
> Hi Steve,
> 
>> On 9 Feb 2017, at 4:09 PM, Steve Wills  wrote:
>>
>> Ports and packages are maintained on the assumption that the user is
>> using a supported version of the OS. We didn't decide when to end
>> support for 10.1 or 10.2. How long after the end of life for 10.1 would
>> you have ports maintain support?
> 
> The issue is quite simple and cause of multiple issues:
> 
> FreeBSD package management makes an ABI promise in the form of
> "FreeBSD:10:amd64", but not even pkg code itself adheres to this,
> and thus we have had subtle and yet fatal breakage in 10.2 and 10.3.
> 
> And since pkg acts according to the imposed paradigm of a unified
> ABI, this will continue to be a source of problems for users who
> cannot know what's going on, lest even know how to fix it.
> 
> They simply run:
> 
> # pkg upgrade
> 
> And then their systems are unusable *and* not fixable with pkg.
> 
> What are they supposed to do?  They come here.  They want to make
> everyone aware that this is a serious issue that shouldn't repeat
> in FreeBSD 11.  It's not to late to do that by pinning the pkg ABI
> to 11.0 by forcing the proper feature macros.

They're supposed to upgrade to a supported version of FreeBSD.

Steve




signature.asc
Description: OpenPGP digital signature

Re: Install of pkg fuse-ntfs fails because of undefined symbol in pkg!?!

[Franco Fichtner]

Hi Steve,

> On 9 Feb 2017, at 4:09 PM, Steve Wills  wrote:
> 
> Ports and packages are maintained on the assumption that the user is
> using a supported version of the OS. We didn't decide when to end
> support for 10.1 or 10.2. How long after the end of life for 10.1 would
> you have ports maintain support?

The issue is quite simple and cause of multiple issues:

FreeBSD package management makes an ABI promise in the form of
"FreeBSD:10:amd64", but not even pkg code itself adheres to this,
and thus we have had subtle and yet fatal breakage in 10.2 and 10.3.

And since pkg acts according to the imposed paradigm of a unified
ABI, this will continue to be a source of problems for users who
cannot know what's going on, lest even know how to fix it.

They simply run:

# pkg upgrade

And then their systems are unusable *and* not fixable with pkg.

What are they supposed to do?  They come here.  They want to make
everyone aware that this is a serious issue that shouldn't repeat
in FreeBSD 11.  It's not to late to do that by pinning the pkg ABI
to 11.0 by forcing the proper feature macros.


Cheers,
Franco
___
freebsd-ports@freebsd.org mailing list
https://lists.freebsd.org/mailman/listinfo/freebsd-ports
To unsubscribe, send any mail to "freebsd-ports-unsubscr...@freebsd.org"

Re: Install of pkg fuse-ntfs fails because of undefined symbol in pkg!?!

[Steve Wills]

Hi,

On 02/09/2017 03:55, Franco Fichtner wrote:
> 
>> On 9 Feb 2017, at 9:49 AM, Kirill Ponomarev  wrote:
>>
>> I don't understand all critics I see in this thread and in your mail,
>> the fate of this project is all in your hands - try to contribute more,
> 
> I'm going to stop you right there.
> 
> That's not entirely true.  Too few committers, substantial lack of
> maintainer feedback, apparently no mentors to pick up newcomers, and
> worst of all no newcomers.

I'm currently mentoring or co-mentoring 5 people. I've done everything
else I can to encourage others to mentor, I think. If you have
suggestions of someone who needs mentoring, let us know and we can try
to find a mentor.

There's nothing we can do about lack of maintainer feedback except reset
maintainers. But keep in mind that lack of maintainer feedback shouldn't
be an issue, we can commit changes after maintainer feedback timeout.

> Decide for yourself which of these are true and which are self-made
> due to project management policies.

What policies do you think need to be changed and in what specific ways?

Steve



signature.asc
Description: OpenPGP digital signature

Re: Install of pkg fuse-ntfs fails because of undefined symbol in pkg!?!

[Steve Wills]

Hi,

On 02/08/2017 12:34, scratch65...@att.net wrote:
> 
> I *did* check for bug reports.  I did a search on "utimenstat"
> and found exactly one, which had been withdrawn as not being a
> bug. 
> 
> But it *is* a bug.  It's a bug on several levels, the most
> significant of which is that the overly frantic schedule makes
> versions have the lifespan of a mayfly.   And we're told "just
> upgrade", as though there's some physical law mandating the
> craziness.

Ports and packages are maintained on the assumption that the user is
using a supported version of the OS. We didn't decide when to end
support for 10.1 or 10.2. How long after the end of life for 10.1 would
you have ports maintain support?

> There are people for whom the system is a tool, not a hobby. They
> don't want to have to rebuild their tools any more than
> carpenters want to replace their hammers and levels every year or
> two.  

If you've having trouble upgrading that are causing you to rebuild, then
that's a different issue, but not one I can help with. It doesn't change
the fact that we don't support unsupported versions of the OS.

> For those people (I'm one) long version lifespans and bug-free
> operation  is a much bigger desideratum than winning the secret
> race (I presume there is some kind of secret race going on, since
> otherwise the crazy scheduling makes No Sense At All).  I can't
> work out what the strategy for winning is, if there is a
> strategy, but I do know that it's not working.   Linux has all
> but won already, and that's sickening.

Ports are maintained by volunteers. If you would like to volunteer to
support branches for longer periods of time, let's talk about that.

> I've been using the o/s since before v2 (I still have the cds)
> and have watched FreeBSD go from being the leading Unix on Intel
> boxes to all-but-dead.  I don't know how to express how saddened
> I feel about that.

I think ports are really improving and the rate of improvement is going up.

Steve



signature.asc
Description: OpenPGP digital signature

Re: Install of pkg fuse-ntfs fails because of undefined symbol in pkg!?!

[Franco Fichtner]

> On 9 Feb 2017, at 10:03 AM, Kirill Ponomarev  wrote:
> 
> On 02/09, Franco Fichtner wrote:
>> 
>>> On 9 Feb 2017, at 9:49 AM, Kirill Ponomarev  wrote:
>>> 
>>> I don't understand all critics I see in this thread and in your mail,
>>> the fate of this project is all in your hands - try to contribute more,
>> 
>> I'm going to stop you right there.
>> 
>> That's not entirely true.  Too few committers, substantial lack of
>> maintainer feedback, apparently no mentors to pick up newcomers, and
>> worst of all no newcomers.
>> 
>> Decide for yourself which of these are true and which are self-made
>> due to project management policies.
> 
> I think I've dejavu, as I've been reading these statements since 15
> years, and they've the same meaning and context, with the same
> argumentations and facts.  The truth is - the dogs bark, but the
> caravan moves on.

The lack of an open reply is exactly what I was arguing for, so I'm not
sure how much this stance helps your case.  ;)


Cheers,
Franco
___
freebsd-ports@freebsd.org mailing list
https://lists.freebsd.org/mailman/listinfo/freebsd-ports
To unsubscribe, send any mail to "freebsd-ports-unsubscr...@freebsd.org"

Re: Install of pkg fuse-ntfs fails because of undefined symbol in pkg!?!

[Kirill Ponomarev]

On 02/09, Franco Fichtner wrote:
> 
> > On 9 Feb 2017, at 9:49 AM, Kirill Ponomarev  wrote:
> > 
> > I don't understand all critics I see in this thread and in your mail,
> > the fate of this project is all in your hands - try to contribute more,
> 
> I'm going to stop you right there.
> 
> That's not entirely true.  Too few committers, substantial lack of
> maintainer feedback, apparently no mentors to pick up newcomers, and
> worst of all no newcomers.
> 
> Decide for yourself which of these are true and which are self-made
> due to project management policies.

I think I've dejavu, as I've been reading these statements since 15
years, and they've the same meaning and context, with the same
argumentations and facts.  The truth is - the dogs bark, but the
caravan moves on.

K.


signature.asc
Description: PGP signature

Re: Install of pkg fuse-ntfs fails because of undefined symbol in pkg!?!

[Franco Fichtner]

> On 9 Feb 2017, at 9:49 AM, Kirill Ponomarev  wrote:
> 
> I don't understand all critics I see in this thread and in your mail,
> the fate of this project is all in your hands - try to contribute more,

I'm going to stop you right there.

That's not entirely true.  Too few committers, substantial lack of
maintainer feedback, apparently no mentors to pick up newcomers, and
worst of all no newcomers.

Decide for yourself which of these are true and which are self-made
due to project management policies.


Cheers,
Franco
___
freebsd-ports@freebsd.org mailing list
https://lists.freebsd.org/mailman/listinfo/freebsd-ports
To unsubscribe, send any mail to "freebsd-ports-unsubscr...@freebsd.org"

Re: Install of pkg fuse-ntfs fails because of undefined symbol in pkg!?!

[Kirill Ponomarev]

On 02/08, list-freebsd-po...@jyborn.se wrote:
> On Wed, Feb 08, 2017 at 12:34:36PM -0500, scratch65...@att.net wrote:
> > For those people (I'm one) long version lifespans and bug-free
> > operation  is a much bigger desideratum than winning the secret
> > race (I presume there is some kind of secret race going on, since
> > otherwise the crazy scheduling makes No Sense At All).  I can't
> > work out what the strategy for winning is, if there is a
> > strategy, but I do know that it's not working.   Linux has all
> > but won already, and that's sickening.
> > 
> > I've been using the o/s since before v2 (I still have the cds)
> > and have watched FreeBSD go from being the leading Unix on Intel
> > boxes to all-but-dead.  I don't know how to express how saddened
> > I feel about that.
> 
> My feelings exactly. I've been a FreeBSD user and strong advocater
> since around FreeBSD v1/v2. But the last few years the management
> of FreeBSD has steered away from making the best server OS, and
> instead focusing on ... what exactly?

Your statement is based on what?

> Making the ports system capable of handling a totally overwhelming
> number of more or less meaningless ports of different versions and
> flavours, and rolling/retiring the base releases at high speed just
> to avoid drowning under the workload of keeping all those ports
> functional?
> 
> Who needs/wants this evolution in a server OS?
> (Linux already owns the desktop, let's not waste any time
> discussing that regrettable fact.)
> 
> I'm sorry, and apologies to all great heroes doing all the
> volunteer work on both FreeBSD base and the ports system,
> but this is how I feel about what is going on with FreeBSD.
> 
> As I have never contributed to FreeBSD in any way, other than
> promoting it to others, I don't have a say in the matter, nor
> do I expect anyone to care about my view. But I'm just really,
> really sad to follow what in my opinion is the slow demise of
> FreeBSD.
> 
> Sorry about all the negativism.

I don't understand all critics I see in this thread and in your mail,
the fate of this project is all in your hands - try to contribute more,
collaborate with developers more, give your ideas and proposals, think
about its future and try to improve it, particiapte in FreeBSD
conferences and workshops, read code and code.

Rephrasing famous quote - if you judge something, you have no time to
love it.

K.


signature.asc
Description: PGP signature

Re: Install of pkg fuse-ntfs fails because of undefined symbol in pkg!?!

[list-freebsd-ports]

On Wed, Feb 08, 2017 at 12:34:36PM -0500, scratch65...@att.net wrote:
> For those people (I'm one) long version lifespans and bug-free
> operation  is a much bigger desideratum than winning the secret
> race (I presume there is some kind of secret race going on, since
> otherwise the crazy scheduling makes No Sense At All).  I can't
> work out what the strategy for winning is, if there is a
> strategy, but I do know that it's not working.   Linux has all
> but won already, and that's sickening.
> 
> I've been using the o/s since before v2 (I still have the cds)
> and have watched FreeBSD go from being the leading Unix on Intel
> boxes to all-but-dead.  I don't know how to express how saddened
> I feel about that.

My feelings exactly. I've been a FreeBSD user and strong advocater
since around FreeBSD v1/v2. But the last few years the management
of FreeBSD has steered away from making the best server OS, and
instead focusing on ... what exactly?

Making the ports system capable of handling a totally overwhelming
number of more or less meaningless ports of different versions and
flavours, and rolling/retiring the base releases at high speed just
to avoid drowning under the workload of keeping all those ports
functional?

Who needs/wants this evolution in a server OS?
(Linux already owns the desktop, let's not waste any time
discussing that regrettable fact.)

I'm sorry, and apologies to all great heroes doing all the
volunteer work on both FreeBSD base and the ports system,
but this is how I feel about what is going on with FreeBSD.

As I have never contributed to FreeBSD in any way, other than
promoting it to others, I don't have a say in the matter, nor
do I expect anyone to care about my view. But I'm just really,
really sad to follow what in my opinion is the slow demise of
FreeBSD.

Sorry about all the negativism.

Peter
___
freebsd-ports@freebsd.org mailing list
https://lists.freebsd.org/mailman/listinfo/freebsd-ports
To unsubscribe, send any mail to "freebsd-ports-unsubscr...@freebsd.org"

Re: Install of pkg fuse-ntfs fails because of undefined symbol in pkg!?!

[scratch65535]

On Wed, 8 Feb 2017 14:43:25 +, Matthew Seaman
 wrote:

>On 02/08/17 11:56, scratch65...@att.net wrote:
>> So, what's the deal here?  To "encourage" people to upgrade, pkg
>> will break their existing install?  That is both hostile and
>> deeply arrogant!
>
>mat's response is more exasperation than anything else.  This is a well
>known problem for which there have been /numerous/ bug reports.  You're
>meant to check that there isn't already a relevant bug report when
>creating a PR in bugzilla.

I *did* check for bug reports.  I did a search on "utimenstat"
and found exactly one, which had been withdrawn as not being a
bug. 

But it *is* a bug.  It's a bug on several levels, the most
significant of which is that the overly frantic schedule makes
versions have the lifespan of a mayfly.   And we're told "just
upgrade", as though there's some physical law mandating the
craziness.

There are people for whom the system is a tool, not a hobby. They
don't want to have to rebuild their tools any more than
carpenters want to replace their hammers and levels every year or
two.  

For those people (I'm one) long version lifespans and bug-free
operation  is a much bigger desideratum than winning the secret
race (I presume there is some kind of secret race going on, since
otherwise the crazy scheduling makes No Sense At All).  I can't
work out what the strategy for winning is, if there is a
strategy, but I do know that it's not working.   Linux has all
but won already, and that's sickening.

I've been using the o/s since before v2 (I still have the cds)
and have watched FreeBSD go from being the leading Unix on Intel
boxes to all-but-dead.  I don't know how to express how saddened
I feel about that.
___
freebsd-ports@freebsd.org mailing list
https://lists.freebsd.org/mailman/listinfo/freebsd-ports
To unsubscribe, send any mail to "freebsd-ports-unsubscr...@freebsd.org"

Re: Install of pkg fuse-ntfs fails because of undefined symbol in pkg!?!

[Matthew Seaman]

On 02/08/17 11:56, scratch65...@att.net wrote:
> So, what's the deal here?  To "encourage" people to upgrade, pkg
> will break their existing install?  That is both hostile and
> deeply arrogant!

mat's response is more exasperation than anything else.  This is a well
known problem for which there have been /numerous/ bug reports.  You're
meant to check that there isn't already a relevant bug report when
creating a PR in bugzilla.

Anyhow:

Your installation is *not* broken.  If you build net-mgmt/pkg (yes, the
latest one) from sources using the ports, you will find it works
perfectly well.

Or you can simply use pkg-static

However, there is no guarantee that

  a) some other package will not suffer from the same utimensat problem
 (so be very wary of using the pre-compiled FreeBSD packages)

  b) or that the package you want will compile properly on 10.2-RELEASE
 (most will, but a number of packages will fall foul of compiler
  problems and the like which have been solved in 10.3-RELEASE.)

Yes, pkg(8) should issue prominent warnings when used on a system that
is out of support.  However, your best recourse is to update your
system, and then all of this grief will cease to bother you.

Cheers,

Matthew




signature.asc
Description: OpenPGP digital signature

Re: Install of pkg fuse-ntfs fails because of undefined symbol in pkg!?!

[scratch65535]

[Default] On Wed, 8 Feb 2017 12:46:09 +0100, Franco Fichtner
 wrote:

>
>> On 8 Feb 2017, at 12:29 PM,   
>> wrote:
>> 
>> I just tried to install the fuse-nfts pkg under 10.2 on my
>> server-of-all-work.  But after requiring me to "upgrade" pkg, the
>> fuse-ntfs install failed, apparently because there's an undefined
>> symbol ("utimenstat") in pkg itself!
>> 
>> How do I extricate myself?
>
>The latest supported release is FreeBSD 10.3 and packages are therefore
>build against it.  It creates this soft breakage inside the fixed ABI,
>which quite a few people run into.
>
>There are two solutions:
>
>(a) Build the packages yourself on a FreeBSD 10.2.
>
>(b) Upgrade to FreeBSD 10.3 and do a "pkg upgrade -f" run.

Perhaps it's just that I've spent too much of my life doing
human-factors work, but if pkg doesn't want to do anything for me
once my version has passed its use-by date, it damned well
shouldn't do ANYTHING.  It should just tell me "nope, 10.2 is too
old, entirely too oll.  No packages for you!"  

The one thing it should NEVER do is break its existing install!  

I agree with your urging, Franco, that the problem be solved in a
general and non-hostile way.

Now, how do I revert the pkg version? :-(
___
freebsd-ports@freebsd.org mailing list
https://lists.freebsd.org/mailman/listinfo/freebsd-ports
To unsubscribe, send any mail to "freebsd-ports-unsubscr...@freebsd.org"

Re: Install of pkg fuse-ntfs fails because of undefined symbol in pkg!?!

[scratch65535]

My bug report just got closed by Mathieu Arnold because "You are
using an obsolete FreeBSD version, you need to update to 10.3."

So, what's the deal here?  To "encourage" people to upgrade, pkg
will break their existing install?  That is both hostile and
deeply arrogant!

I don't want to upgrade because (a) I'm perfectly happy with 10.2
and (b) I've never been able to upgrade in place.  Every attempt
has ended in my having scour everything down to the metal and
reinstall.  I'm not willing to waste days doing that for no good
reason.
___
freebsd-ports@freebsd.org mailing list
https://lists.freebsd.org/mailman/listinfo/freebsd-ports
To unsubscribe, send any mail to "freebsd-ports-unsubscr...@freebsd.org"

Re: Install of pkg fuse-ntfs fails because of undefined symbol in pkg!?!

[Franco Fichtner]

> On 8 Feb 2017, at 12:29 PM,   
> wrote:
> 
> I just tried to install the fuse-nfts pkg under 10.2 on my
> server-of-all-work.  But after requiring me to "upgrade" pkg, the
> fuse-ntfs install failed, apparently because there's an undefined
> symbol ("utimenstat") in pkg itself!
> 
> How do I extricate myself?

The latest supported release is FreeBSD 10.3 and packages are therefore
build against it.  It creates this soft breakage inside the fixed ABI,
which quite a few people run into.

There are two solutions:

(a) Build the packages yourself on a FreeBSD 10.2.

(b) Upgrade to FreeBSD 10.3 and do a "pkg upgrade -f" run.

And for the wicked:

(c) Let's please address this issue within FreeBSD so users don't run into it.

;)


Cheers,
Franco
___
freebsd-ports@freebsd.org mailing list
https://lists.freebsd.org/mailman/listinfo/freebsd-ports
To unsubscribe, send any mail to "freebsd-ports-unsubscr...@freebsd.org"