Re: [gentoo-user] Question about initial/default ownership of /usr/portage

[Alan McKinnon]

On 03/10/2015 10:26, Neil Bothwick wrote:
> On Fri, 2 Oct 2015 20:48:06 -0700, Raymond Jennings wrote:
> 
>> Who is supposed to own /usr/portage?
> 
> This was discussed in some detail two weeks ago. Search for the thread
> entitled "portage directory ownerships?".
> 
> 


There's an aspect of that we didn't discuss 2 weeks ago.

The answer to the OP's question is, as always, "it depends".
>From man 5 make.conf, section "FEATURES":

  userfetch
 When  portage  is  run  as root, drop privileges to
 portage:portage  during  the  fetching  of  package
 sources.

  userpriv
 Allow  portage  to drop root privileges and compile
 packages  as  portage:portage  without  a   sandbox
 (unless usersandbox is also used).

  usersandbox
 Enable  the sandbox in the compile phase, when run‐
 ning without root privs (userpriv).

  usersync
 Drop privileges to the owner of  ${repository_loca‐
 tion}  for  emerge(1)  --sync operations. Note that
 this feature assumes  that  all  subdirectories  of
 ${repository_location}  have  the same ownership as
 ${repository_location} itself.  It  is  the  user's
 responsibility  to  ensure correct ownership, since
 otherwise Portage would have to waste time validat‐
 ing ownership for each and every sync operation.


So logically, if user* is in FEATURES then $PORTDIR should be
portage:portage
If emerge --sync is done as user alan, then all repos ($PORTDIR, local
overlays, layman overlays) need to be owned by alan:

if user* is not in FEATURES and everything gets run as root, then
PORTDIR, DISTDIR and so on can be root:root.

The point is, there really isn't a "owner:group /should/ be" rule for
portage data, the admin needs to make it whatever he needs it to be.

-- 
Alan McKinnon
alan.mckin...@gmail.com

Re: [gentoo-user] Question about initial/default ownership of /usr/portage

[Neil Bothwick]

On Sat, 3 Oct 2015 10:40:42 +0200, Alan McKinnon wrote:

> On 03/10/2015 10:26, Neil Bothwick wrote:
> > On Fri, 2 Oct 2015 20:48:06 -0700, Raymond Jennings wrote:
> >   
> >> Who is supposed to own /usr/portage?  
> > 
> > This was discussed in some detail two weeks ago. Search for the thread
> > entitled "portage directory ownerships?".
> > 
> >   
> 
> 
> There's an aspect of that we didn't discuss 2 weeks ago.
> 
> The answer to the OP's question is, as always, "it depends".
> From man 5 make.conf, section "FEATURES":
> 
>   userfetch
>  When  portage  is  run  as root, drop privileges to
>  portage:portage  during  the  fetching  of  package
>  sources.
> 
>   userpriv
>  Allow  portage  to drop root privileges and compile
>  packages  as  portage:portage  without  a   sandbox
>  (unless usersandbox is also used).
> 
>   usersandbox
>  Enable  the sandbox in the compile phase, when run‐
>  ning without root privs (userpriv).
> 
>   usersync
>  Drop privileges to the owner of  ${repository_loca‐
>  tion}  for  emerge(1)  --sync operations. Note that
>  this feature assumes  that  all  subdirectories  of
>  ${repository_location}  have  the same ownership as
>  ${repository_location} itself.  It  is  the  user's
>  responsibility  to  ensure correct ownership, since
>  otherwise Portage would have to waste time validat‐
>  ing ownership for each and every sync operation.
> 
> 
> So logically, if user* is in FEATURES then $PORTDIR should be
> portage:portage
> If emerge --sync is done as user alan, then all repos ($PORTDIR, local
> overlays, layman overlays) need to be owned by alan:
> 
> if user* is not in FEATURES and everything gets run as root, then
> PORTDIR, DISTDIR and so on can be root:root.

That's interesting, but I think only the last one applies to $POSTDIR.
userfetch applies to $DISTDIR, which may or may not be within $PORTDIR
while the others relate to compiling so should only affect transient
directories in $PORTAGE_TMPDIR.

> The point is, there really isn't a "owner:group /should/ be" rule for
> portage data, the admin needs to make it whatever he needs it to be.

The default though is for all four of those features to be set, so
you would expect $PORTDIR to be portage:portage, yet on this recent
install it is root:root. The local portage mirror it syncs from is
portage:portage.

I'm beginning to think portage is female and all attempts to understand it
are futile...


-- 
Neil Bothwick

Give a man a fish and you feed him for a day; teach him
to use the Net and he won't bother you for weeks.


pgpd9yXAapzxZ.pgp
Description: OpenPGP digital signature

Re: [gentoo-user] FF increasingly being bundled up

[Dale]

Alec Ten Harmsel wrote:
> On Sat, Oct 03, 2015 at 10:15:08AM +0100, Mick wrote:
>> Hi All,
>>
>> I happened to notice that Firefox is having more and more USE flags for 
>> using 
>> system-side rather than bundled in libraries:
>>
>> ...
>>
>> However, such libraries are not enabled by default and FF will be 
>> downloading 
>> 177,578 KiB of sources.  A couple of questions:
>>
>> - Will FF be downloading less if the system side USE flags are enabled?
> As far as I can tell, the same tarball is downloaded no matter which
> system-* USE flags you set. The ebuild tells the build system to use the
> system stuff.
>
>
> Alec
>
>


That's true.  Regardless of USE flags, same tarball.  I don't recall
ever seeing USE flags affect the size of the tarball unless it is a
binary which likely doesn't have USE flags anyway. 

Dale

:-)  :-) 

Re: [gentoo-user] Hard drive click of death?

[Dale]

Mick wrote:
> On Saturday 03 Oct 2015 12:23:47 Dale wrote:
>> Mick wrote:
>>> Hi All,
>>>
>>> Recently I noticed that an old drive is clicking 3 or 4 times when it is
>>> powered up.  Thereafter is stays quiet.  I think it is a mechanical
>>> fault, because I placed the box on its side and the drive did not click
>>> when booting up.
>>>
>>> I ran smartctl short/long/conveyance tests and no errors are reported.
>>>
>>> In your experience is this something to concern myself with?  There is
>>> not business critical data on the drive at present.
>> I have found that if SMART reports a error, it is good to replace it as
>> soon as you can.  When a drive makes a noise that isn't normal, that's
>> also a sign that you need to replace it.  If you google around for that
>> model of drive, you may can find where others have had the same and they
>> shed some light on what happened, it died, it ran for ages and is normal
>> or something else.
>>
>> Right now, backups would be a good idea.  Doing some drive shopping
>> would to unless google turns up something that says it is nothing to
>> worry about, doubtful tho.
>>
>> Dale
>>
>> :-)  :-)
> I don't know if I am getting more nostalgic in my old days, but it is things 
> like this that make me like spinning drives.  More often than not they give 
> some kind of warning.  :-)
>
> That said I've not yet had an SSD going sideways on me.
>


I plan to get me a SSD one of these days.  I'm going to let them work
out the kinks first tho, at someone else's expense that can afford the
things. 

At least bearings do give a warning.  The SMART tool can give early
warnings too.  I have had SMART save my data twice now. 

Dale

:-)  :-) 

[gentoo-user] FF increasingly being bundled up

[Mick]

Hi All,

I happened to notice that Firefox is having more and more USE flags for using 
system-side rather than bundled in libraries:

[ebuild U  ] www-client/firefox-38.3.0::gentoo [38.2.1::gentoo] USE="dbus 
gmp-autoupdate jemalloc3 jit minimal -bindist -custom-cflags -custom-
optimization -debug -egl -gstreamer -gstreamer-0 -hardened (-neon) (-pgo) -
pulseaudio (-selinux) -startup-notification -system-cairo -system-icu -system-
jpeg -system-libvpx -system-sqlite {-test} -wifi"


However, such libraries are not enabled by default and FF will be downloading 
177,578 KiB of sources.  A couple of questions:

- Will FF be downloading less if the system side USE flags are enabled?
- What are the benefits or disbenefits of enabling them?

-- 
Regards,
Mick


signature.asc
Description: This is a digitally signed message part.

Re: [gentoo-user] Hard drive click of death?

[Alan McKinnon]

On 03/10/2015 11:36, Mick wrote:
> Hi All,
> 
> Recently I noticed that an old drive is clicking 3 or 4 times when it is 
> powered up.  Thereafter is stays quiet.  I think it is a mechanical fault, 
> because I placed the box on its side and the drive did not click when booting 
> up.
> 
> I ran smartctl short/long/conveyance tests and no errors are reported.
> 
> In your experience is this something to concern myself with?  There is not 
> business critical data on the drive at present.


Hard to tell. Obviously you have mechanical wear as the click stops when
the drive is tilted. That indicates slight wear in moving parts.

But for now the drive is doing it's job without errors. So all we can
say is that you probably have a raised likelihood of a real failure
sooner rather than later.

How you approach this depends on your needs and budget. If you have good
backups and can go without the data for a few days, there's no real harm
is waiting till it really packs up. If drives are cheap where you are
you might as well replace it now. I don't see any compelling reason in
your case why you should do one or the other at this point.


-- 
Alan McKinnon
alan.mckin...@gmail.com

Re: [gentoo-user] FF increasingly being bundled up

[Mick]

On Saturday 03 Oct 2015 10:37:07 Marc Joliet wrote:
> On Saturday 03 October 2015 10:15:08 Mick wrote:
> >Hi All,
> >
> >I happened to notice that Firefox is having more and more USE flags for
> >using system-side rather than bundled in libraries:
> >
> >[ebuild U  ] www-client/firefox-38.3.0::gentoo [38.2.1::gentoo]
> >USE="dbus gmp-autoupdate jemalloc3 jit minimal -bindist -custom-cflags
> >-custom- optimization -debug -egl -gstreamer -gstreamer-0 -hardened
> >(-neon) (-pgo) - pulseaudio (-selinux) -startup-notification
> >-system-cairo -system-icu -system- jpeg -system-libvpx -system-sqlite
> >{-test} -wifi"
> >
> >
> >However, such libraries are not enabled by default and FF will be
> >downloading 177,578 KiB of sources.  A couple of questions:
> >
> >- Will FF be downloading less if the system side USE flags are enabled?
> >- What are the benefits or disbenefits of enabling them?
> 
> I don't know about the size of the downloads, but is that *really* a
> problem?

When you are on a metered ADSL connection, yes, it can be.


> As to the system-* flags:  I've been setting all of them except for system-
> sqlite since... they exist, I think.  I make the exception to system-sqlite
> because it likely has a negative performance impact on other packages that
> depend on sqlite.  However, it seems that at least system-cairo might be
> problematic in recent versions of Firefox, see
> https://bugs.gentoo.org/show_bug.cgi?id=558150.
> 
> In general, I think it depends on how well upstream supports the flags and
> how quickly problems are noticed when using them, which might not be as
> quick as we like.
> 
> HTH

Have you found that using local system flags causes rebuilds of FF more often?  
I seem to build or rebuild FF every couple of weeks, but I would not want to 
have to rebuild it more often.

-- 
Regards,
Mick


signature.asc
Description: This is a digitally signed message part.

Re: [gentoo-user] Hard drive click of death?

[Dale]

Mick wrote:
> Hi All,
>
> Recently I noticed that an old drive is clicking 3 or 4 times when it is 
> powered up.  Thereafter is stays quiet.  I think it is a mechanical fault, 
> because I placed the box on its side and the drive did not click when booting 
> up.
>
> I ran smartctl short/long/conveyance tests and no errors are reported.
>
> In your experience is this something to concern myself with?  There is not 
> business critical data on the drive at present.


I have found that if SMART reports a error, it is good to replace it as
soon as you can.  When a drive makes a noise that isn't normal, that's
also a sign that you need to replace it.  If you google around for that
model of drive, you may can find where others have had the same and they
shed some light on what happened, it died, it ran for ages and is normal
or something else. 

Right now, backups would be a good idea.  Doing some drive shopping
would to unless google turns up something that says it is nothing to
worry about, doubtful tho. 

Dale

:-)  :-) 

Re: [gentoo-user] System76 Hardware

[Nuno Magalhães]

On Fri, Oct 2, 2015 at 9:55 AM, Emanuele Rusconi  wrote:
>> [1] http://www.pcidatabase.com/
>
>
> I didn't know that. It doesn't seem to have System76 in the database,
> though.

That's because System76 doesn't manufacture hardware components, but
(at least most of) those found in one of their laptops will likely be
in that database.

Re: [gentoo-user] FF increasingly being bundled up

[Alec Ten Harmsel]

On Sat, Oct 03, 2015 at 10:57:47AM +0100, Mick wrote:
> 
> Have you found that using local system flags causes rebuilds of FF more 
> often?  
> I seem to build or rebuild FF every couple of weeks, but I would not want to 
> have to rebuild it more often.
> 

I have been using all the system-* flags as long as I can remember. I
can't remember the last time that I rebuilt firefox for a reason other
than a version update or I had directly changed its USE flags. I usually
update weekly, so I see firefox rebuilds probably once a month or a
little less frequently on average.

Alec

Re: [gentoo-user] Hard drive click of death?

[Mick]

On Saturday 03 Oct 2015 12:23:47 Dale wrote:
> Mick wrote:
> > Hi All,
> > 
> > Recently I noticed that an old drive is clicking 3 or 4 times when it is
> > powered up.  Thereafter is stays quiet.  I think it is a mechanical
> > fault, because I placed the box on its side and the drive did not click
> > when booting up.
> > 
> > I ran smartctl short/long/conveyance tests and no errors are reported.
> > 
> > In your experience is this something to concern myself with?  There is
> > not business critical data on the drive at present.
> 
> I have found that if SMART reports a error, it is good to replace it as
> soon as you can.  When a drive makes a noise that isn't normal, that's
> also a sign that you need to replace it.  If you google around for that
> model of drive, you may can find where others have had the same and they
> shed some light on what happened, it died, it ran for ages and is normal
> or something else.
> 
> Right now, backups would be a good idea.  Doing some drive shopping
> would to unless google turns up something that says it is nothing to
> worry about, doubtful tho.
> 
> Dale
> 
> :-)  :-)

I don't know if I am getting more nostalgic in my old days, but it is things 
like this that make me like spinning drives.  More often than not they give 
some kind of warning.  :-)

That said I've not yet had an SSD going sideways on me.

-- 
Regards,
Mick


signature.asc
Description: This is a digitally signed message part.

Re: [gentoo-user] FF increasingly being bundled up

[Marc Joliet]

On Saturday 03 October 2015 10:15:08 Mick wrote:
>Hi All,
>
>I happened to notice that Firefox is having more and more USE flags for using
>system-side rather than bundled in libraries:
>
>[ebuild U  ] www-client/firefox-38.3.0::gentoo [38.2.1::gentoo] USE="dbus
>gmp-autoupdate jemalloc3 jit minimal -bindist -custom-cflags -custom-
>optimization -debug -egl -gstreamer -gstreamer-0 -hardened (-neon) (-pgo) -
>pulseaudio (-selinux) -startup-notification -system-cairo -system-icu
>-system- jpeg -system-libvpx -system-sqlite {-test} -wifi"
>
>
>However, such libraries are not enabled by default and FF will be downloading
>177,578 KiB of sources.  A couple of questions:
>
>- Will FF be downloading less if the system side USE flags are enabled?
>- What are the benefits or disbenefits of enabling them?

I don't know about the size of the downloads, but is that *really* a problem?  

As to the system-* flags:  I've been setting all of them except for system-
sqlite since... they exist, I think.  I make the exception to system-sqlite 
because it likely has a negative performance impact on other packages that 
depend on sqlite.  However, it seems that at least system-cairo might be 
problematic in recent versions of Firefox, see 
https://bugs.gentoo.org/show_bug.cgi?id=558150.

In general, I think it depends on how well upstream supports the flags and how 
quickly problems are noticed when using them, which might not be as quick as 
we like.

HTH
-- 
Marc Joliet
--
"People who think they know everything really annoy those of us who know we
don't" - Bjarne Stroustrup


signature.asc
Description: This is a digitally signed message part.

[gentoo-user] Hard drive click of death?

[Mick]

Hi All,

Recently I noticed that an old drive is clicking 3 or 4 times when it is 
powered up.  Thereafter is stays quiet.  I think it is a mechanical fault, 
because I placed the box on its side and the drive did not click when booting 
up.

I ran smartctl short/long/conveyance tests and no errors are reported.

In your experience is this something to concern myself with?  There is not 
business critical data on the drive at present.
-- 
Regards,
Mick


signature.asc
Description: This is a digitally signed message part.

Re: [gentoo-user] FF increasingly being bundled up

[Bryan Gardiner]

On Sat, Oct 03, 2015 at 06:26:37AM -0500, Dale wrote:
> Alec Ten Harmsel wrote:
> > On Sat, Oct 03, 2015 at 10:15:08AM +0100, Mick wrote:
> >> Hi All,
> >>
> >> I happened to notice that Firefox is having more and more USE flags for 
> >> using 
> >> system-side rather than bundled in libraries:
> >>
> >> ...
> >>
> >> However, such libraries are not enabled by default and FF will be 
> >> downloading 
> >> 177,578 KiB of sources.  A couple of questions:
> >>
> >> - Will FF be downloading less if the system side USE flags are enabled?
> > As far as I can tell, the same tarball is downloaded no matter which
> > system-* USE flags you set. The ebuild tells the build system to use the
> > system stuff.
> 
> That's true.  Regardless of USE flags, same tarball.  I don't recall
> ever seeing USE flags affect the size of the tarball unless it is a
> binary which likely doesn't have USE flags anyway. 

"emerge -pf" shows what distfiles are required and takes USE flags
into account, if you want to check for yourself.

Cheers,
Bryan


signature.asc
Description: Digital signature

Re: [gentoo-user] FF increasingly being bundled up

[Mick]

On Saturday 03 Oct 2015 12:18:19 Alec Ten Harmsel wrote:
> On Sat, Oct 03, 2015 at 10:57:47AM +0100, Mick wrote:
> > Have you found that using local system flags causes rebuilds of FF more
> > often? I seem to build or rebuild FF every couple of weeks, but I would
> > not want to have to rebuild it more often.
> 
> I have been using all the system-* flags as long as I can remember. I
> can't remember the last time that I rebuilt firefox for a reason other
> than a version update or I had directly changed its USE flags. I usually
> update weekly, so I see firefox rebuilds probably once a month or a
> little less frequently on average.
> 
> Alec

Thank you all for your replies.  I've enabled the system-* flags and firefox 
built in the same time give or take a few seconds.  I can't see a difference 
yet, although it feels faster - clear psychological advantage!  ;-)

-- 
Regards,
Mick


signature.asc
Description: This is a digitally signed message part.

Re: [gentoo-user] Question about initial/default ownership of /usr/portage

[Dale]

Neil Bothwick wrote:
> I'm beginning to think portage is female and all attempts to
> understand it are futile... 


You to huh?  Every time I think I got something figured out, it throws a
curve ball at me and hits me in the forehead.  I don't like eating dirt.

o_O

Dale

:-)  :-) 

Re: [gentoo-user] FF increasingly being bundled up

[Alec Ten Harmsel]

On Sat, Oct 03, 2015 at 10:15:08AM +0100, Mick wrote:
> Hi All,
> 
> I happened to notice that Firefox is having more and more USE flags for using 
> system-side rather than bundled in libraries:
> 
> ...
>
> However, such libraries are not enabled by default and FF will be downloading 
> 177,578 KiB of sources.  A couple of questions:
> 
> - Will FF be downloading less if the system side USE flags are enabled?

As far as I can tell, the same tarball is downloaded no matter which
system-* USE flags you set. The ebuild tells the build system to use the
system stuff.

> - What are the benefits or disbenefits of enabling them?

Enabling the use of system libraries could make Firefox unstable, since
the FF developers may have patched the bundled libs. It will also save
compile time, although compared to FF itself, most of the libs that can
be prevented from bundling have short compile times.

For me, personally, disabling those USE flags and using bundled libs
represents some sort of idealogical loss since I view bundling libraries
as a cop out solution. This may or may not matter to you.

Alec

Re: [gentoo-user] FF increasingly being bundled up

[Dale]

Bryan Gardiner wrote:
> On Sat, Oct 03, 2015 at 06:26:37AM -0500, Dale wrote:
>> Alec Ten Harmsel wrote:
>>> On Sat, Oct 03, 2015 at 10:15:08AM +0100, Mick wrote:
 Hi All,

 I happened to notice that Firefox is having more and more USE flags for 
 using 
 system-side rather than bundled in libraries:

 ...

 However, such libraries are not enabled by default and FF will be 
 downloading 
 177,578 KiB of sources.  A couple of questions:

 - Will FF be downloading less if the system side USE flags are enabled?
>>> As far as I can tell, the same tarball is downloaded no matter which
>>> system-* USE flags you set. The ebuild tells the build system to use the
>>> system stuff.
>> That's true.  Regardless of USE flags, same tarball.  I don't recall
>> ever seeing USE flags affect the size of the tarball unless it is a
>> binary which likely doesn't have USE flags anyway. 
> "emerge -pf" shows what distfiles are required and takes USE flags
> into account, if you want to check for yourself.
>
> Cheers,
> Bryan


But the Firefox tarball will be the same size.  If you change a USE
flag, that may pull in dependencies but it doesn't change the size of
the Firefox tarball itself.   The Firefox tarball is getting rather
large.  I'm not sure if that is a good thing or not. 

Dale

:-)  :-) 

Re: [gentoo-user] FF increasingly being bundled up

[Rich Freeman]

On Sat, Oct 3, 2015 at 9:41 AM, Mick  wrote:
>
> Thank you all for your replies.  I've enabled the system-* flags and firefox
> built in the same time give or take a few seconds.  I can't see a difference
> yet, although it feels faster - clear psychological advantage!  ;-)
>

The same tarball is used either way.  We'd need to rip it apart and
distribute modified tarballs ourselves if we wanted to change that,
and this generally isn't how we do things.  In the case of firefox it
might even have legal issues due to how they treat their trademarks,
but I'd have to look into that.

I wouldn't expect using system sqllite to have any negative impact on
other system packages using sqllite.  It might or might not affect
firefox performance if its sqllite build options are different from
the system library.  I believe some sqllite behavior settings are only
configurable at build time.

Besides saving build time, using system libraries would save RAM and
load time as well, since firefox would share those libraries with
other programs vs loading private copies of them.  There is also the
security matter, which is normally a very important reason to use
system libraries, but firefox is probably at least as rigorous about
keeping their bundled libraries updated as Gentoo is, so that could go
either way.

Obviously you're going to have less upstream support in general for
system libraries, but most distros are going to use them anyway.

I'm not quite sure why more of them aren't used by default on Gentoo.
Chromium tries to use as many system libs as they can, with more
experimental ones being defaulted to bundled, and ones that they
haven't managed to unbundle may not be optional.  Google loves to
patch the stuff they distribute, so you can't just switch to system
libs in all cases without fixing all the APIs.

-- 
Rich

Re: [gentoo-user] Hard drive click of death?

[Mick]

On Saturday 03 Oct 2015 10:55:42 Alan McKinnon wrote:
> On 03/10/2015 11:36, Mick wrote:
> > Hi All,
> > 
> > Recently I noticed that an old drive is clicking 3 or 4 times when it is
> > powered up.  Thereafter is stays quiet.  I think it is a mechanical
> > fault, because I placed the box on its side and the drive did not click
> > when booting up.
> > 
> > I ran smartctl short/long/conveyance tests and no errors are reported.
> > 
> > In your experience is this something to concern myself with?  There is
> > not business critical data on the drive at present.
> 
> Hard to tell. Obviously you have mechanical wear as the click stops when
> the drive is tilted. That indicates slight wear in moving parts.
> 
> But for now the drive is doing it's job without errors. So all we can
> say is that you probably have a raised likelihood of a real failure
> sooner rather than later.
> 
> How you approach this depends on your needs and budget. If you have good
> backups and can go without the data for a few days, there's no real harm
> is waiting till it really packs up. If drives are cheap where you are
> you might as well replace it now. I don't see any compelling reason in
> your case why you should do one or the other at this point.

Thanks Alan, I'll take a quick back up of some data I'd like to keep, just in 
case.
-- 
Regards,
Mick


signature.asc
Description: This is a digitally signed message part.

Re: [gentoo-user] FF increasingly being bundled up

[Mick]

On Saturday 03 Oct 2015 15:57:15 Rich Freeman wrote:

> I wouldn't expect using system sqllite to have any negative impact on
> other system packages using sqllite.  It might or might not affect
> firefox performance if its sqllite build options are different from
> the system library.  I believe some sqllite behavior settings are only
> configurable at build time.

Portage asked me to add this USE flag:

 >=dev-db/sqlite-3.8.10.2 secure-delete

> Besides saving build time, using system libraries would save RAM and
> load time as well, since firefox would share those libraries with
> other programs vs loading private copies of them.  

Yes, FF now takes a few milliseconds less to load.  I didn't check the RAM  it 
occupies before and after, but what you say makes sense.
-- 
Regards,
Mick


signature.asc
Description: This is a digitally signed message part.

Re: [gentoo-user] update problems

[lee]

Alan Mackenzie  writes:

> Hello, Lee.
>
> On Tue, Sep 29, 2015 at 08:45:10PM +0200, lee wrote:
>> Neil Bothwick  writes:
>
>> > Patches are always more welcome than suggestions. "Fix it!" is never as
>> > welcome as "here's how". I think it was Canek who said "code talks". 
>
>> Do you have an example for such a case?
>
> Yes, many.  I'm a contributor to Emacs, and relatively frequently (perhaps
> 10 - 30 times a yeaar) somebody reports a bug and simultaneously submits
> a patch for it.  This is always well received,

I sent in a contribution to emacs, too, and never even heard anything
back.

> On the other hand, "wouldn't X be a good idea"s which reach the mailing
> list only rarely get taken up by regular contributors - there's only so
> much time in the day, and such hackers usually have plenty of Xs of
> their own to fill their time with.

That apparently means that nobody is allowed to suggest something and/or
to discuss a suggestion, and that everyone must have an "I don't care"
attitude.

>> My experience has disproved this claim, and I've even seen people
>> fixing stuff multiple times after I told them it's broken and provided
>> a perfectly working version before telling them, much better coded,
>> which they could have used instead of insisting on their crappy code
>> and trying to fix it several times.
>
> That's not very friendly,

How is it not friendly to point out a bug when you find one, at the same
time pointing to what fixes it?

> and hardly inclined to gain extra contributors
> for your project.  A gentle guiding hand, helping these other people to
> reach a satisfactory fix themselves, would work much better.

It wasn't my project but software I'm using and had made a fork of.  So
I noticed what upstream changed, found it to be broken, fixed it and
notified them that it's broken and how, and that there's a fix they can
use.  They didn't use the fix, made a couple attempts to fix their own
code until it finally worked, and though it now works, their code still
sucks.

So the most logical conclusion is not to report bugs and not to provide
any fixes or contributions, and not dare to suggest anything because at
best, it leads to nothing, and most of the time, you're being told that
you're a clueless idiot and to shut up.  OTOH, you often times get to
hear and/or see that peoples' contributions and help are wanted and that
there are always not enough contributers.  But why ask for more
contributers when contributions aren't wanted anyway?

>> > On the contrary, it serves to illustrate that you do not grasp the
>> > complexity of the situation.
>
>> Perhaps you can enlighten me how it is so difficult to change a message
>> from "slot conflict" to "slot conflict (can probably be ignored while
>> there are other problems)" and what the complexity is which makes it
>> impossible to do so.
>
> It's not difficult, it's just tedious.  Something like that which is
> user facing needs to be agreed by the core of the project, and getting
> that agreement tends to involve lots of bike shedding on the project
> mailing lists - there's always a few people who'll prefer the message to
> stay the same.

That is a bad situation which might help to explain why projects neither
want contributions, nor contributers.  Yet it doesn't mean that those
who would like to contribute shall receive the blame for the bad
situation the project is in, nor that it is wise to put them off.

It also indicates that the argument "go ahead and supply a patch" is
entirely inappropriate beyond being merely condescending, and that
arguing along the lines that the contributers aren't being payed and
that /you/ aren't contributing anyway is even worse.  None of them are
acceptable under these conditions.  They are irrelevant.

> Then there's all the stuff about writing change logs for the change
> and commiting it.

How is that being too tedious?  If it really is too tedious, is there a
way to make it less tedious?

> Such a tiny change is scarcely achievable in less than an hour.  To
> the core developers, it barely seems worth it.

So nobody do anything because it isn't worth it.  That's a great
attitude.


-- 
Again we must be afraid of speaking of daemons for fear that daemons
might swallow us.  Finally, this fear has become reasonable.

Re: [gentoo-user] Hard drive click of death?

[Meino . Cramer]

Mick  [15-10-03 21:04]:
> On Saturday 03 Oct 2015 19:44:03 meino.cra...@gmx.de wrote:
> > Mick  [15-10-03 11:44]:
> > > Hi All,
> > > 
> > > Recently I noticed that an old drive is clicking 3 or 4 times when it is
> > > powered up.  Thereafter is stays quiet.  I think it is a mechanical
> > > fault, because I placed the box on its side and the drive did not click
> > > when booting up.
> > > 
> > > I ran smartctl short/long/conveyance tests and no errors are reported.
> > > 
> > > In your experience is this something to concern myself with?  There is
> > > not business critical data on the drive at present.
> > 
> > Hi Mick,
> > 
> > I had a drive doing similiar things whith comparable "features".
> > In my case it was a defective cable, which gave good contact when
> > the drive was put on its side due the slight drill of the cable.
> > But it is less likely as what the other said, though...
> > 
> > HTH!
> > Best regards,
> > Meino
> 
> HA!  I reseated the cable on the drive and the MoBo and tried to start the PC 
> a couple of times.  So far no clicking!  :-)
> 
> Let's hope it stays this way.
> 
> Thank you Meino for your suggestion.
> 
> -- 
> Regards,
> Mick

Hi Mick,

you're welcome! 8)

fingers crossed ! :

HTH! ;)

Best regards,
Meino

Re: [gentoo-user] update problems

[lee]

Neil Bothwick  writes:

> On Tue, 29 Sep 2015 20:45:10 +0200, lee wrote:
>
>> Neil Bothwick  writes:
>> 
>> > Patches are always more welcome than suggestions. "Fix it!" is never
>> > as welcome as "here's how". I think it was Canek who said "code
>> > talks".   
>> 
>> Do you have an example for such a case?
>
> Just look at b.g.o. Many bug reports include a patch submitted by a user
> that makes its way into the tree.

What is b.g.o.?

>> My experience has disproved
>> this claim, and I've even seen people fixing stuff multiple times after
>> I told them it's broken and provided a perfectly working version before
>> telling them, much better coded, which they could have used instead of
>> insisting on their crappy code and trying to fix it several times.
>
> You cannot judge one group of people on the behaviour of an unrelated
> group.

But I can observe the behaviour of many similar groups of people, or of
people, and come to a conclusion about what behaviour can be
expected. That with very few exceptions, neither bug reports, nor fixes
are wanted, is an observation.  I suppose I should have expected the
same behaviour here, and I made the mistake to think that I might
encounter different behaviour here.


> [...]
> #!/usr/lib/python-exec/python-exec2-c
> [...]
>
> In there you will find python2.7/emerge and python3.4/emerge (depending
> on which Python versions you have installed).

ok

>> I don't believe that they let everyone modify what they're working on,
>> so they are the only ones who /can/ fix it.  Besides, show me where I
>> said something like "I want the devs to fix it".
>
> They don't. You submit the modifications in the bug report and they vet
> and apply the patches.

Obviously, no patch is wanted.

>> > Adding the word "just" to a demand does not make the task any
>> > simpler, nor does it increase your chances of getting what you want.  
>> 
>> Go ahead and show me where I have demanded something.
>
> Your insistence that it should be changed amounts to a demand. Your
> assertion that it can be done easily only demeans the efforts of the
> devs, implying that the fix is simple but they cannot be bothered.

I'm not insisting at all.  I'm merely saying that it could easily be
fixed.  So people say it's not easy to fix but incredibly difficult, and
I say that fixing a "print" statement in some script can't be so
incredibly difficult to fix.  Then people agree and give other reasons
--- which have nothing to do with changing a "print" statement --- for
why this is difficult to do.

Some of what they say indicates that the devs cannot be bothered.  How
you conclude that something which could be done easily and isn't demeans
anyones efforts escapes me.  However, you would have to blame the people
saying that the devs cannot be bothered, not me.

>> > On the contrary, it serves to illustrate that you do not grasp the
>> > complexity of the situation.  
>> 
>> Perhaps you can enlighten me how it is so difficult to change a message
>> from "slot conflict" to "slot conflict (can probably be ignored while
>> there are other problems)" and what the complexity is which makes it
>> impossible to do so.
>
> Changing the message is trivial, knowing when to change it is not. Unless
> you can provide a means to tell unimportant slot conflicts from important
> ones, Context is everything and the variety of Gentoo systems out there
> make it extremely difficult for portage to understand the context in
> human terms. 

You don't need to know when to change it.  Once someone finds that they
still cannot update after fixing all other issues, they are able to
figure out that they may not ignore the message.  Apparently it rarely
happens that they may not ignore the message.

Some time, there might be a way to know when to change the message, and
even better messages could be used.  Until then, a small change would go
a long way towards making the system more friendly for the users.  So
why make things difficult for everyone --- for the devs by asking for an
ultimate fix and for the users by giving them misleading messages ---
rather than making things easier by using messages less misleading while
the devs can their time until they find the ultimate fix?

I'd give them credit for taking that step.  Can you explain how taking
such a step, or even suggesting to take it, could demean their efforts?


-- 
Again we must be afraid of speaking of daemons for fear that daemons
might swallow us.  Finally, this fear has become reasonable.

Re: [gentoo-user] black screen on boot when udevevents are processed

[lee]

lee  writes:

>  writes:
>

> [...]
>>> However, I can see the BIOS and the boot manager menu, then during
>>> booting, the screen goes black and the monitor says "no signal"
>>> (probably when the nvidia module is loaded).  So I logged in blindly
>>> and started X11 and got a picture again.
>>> 
>>> The monitor info says 4k@60Hz in the BIOS and when X11 is running.  So
>>> there must be some setting which causes the graphics card to blank out
>>> on the console.
>>> 
>>> Any idea what that could be?
>>
> [...]
>
> So I need to read some documentation tomorrow ...

Does the nvidia module have any options which can specified when it is
loaded?  I haven't found anything about that in the docs.

Using "nomodeswitch" didn't help; blacklisting the module did help.
That proves that loading the module when not starting X11 makes the
screen go blank.


-- 
Again we must be afraid of speaking of daemons for fear that daemons
might swallow us.  Finally, this fear has become reasonable.

[gentoo-user] Gentoo speaking to Android

[Meino . Cramer]

Hi,

my setup is a recent Gentoo system, which is connected to the
internet. This acts as an Wifi hotspot for my android tablet.

I wiresharked some connections of this tablet to the internet
and it was -- lets say -- "astonished", how much this little 
tablet has to tell people, which I dont know and have never heard
of only by astablishing the connection and doing nothing else.

Since rooting an android device is common for nerds and geeks
only most firewalls available for android devices are so called
"no root firewalls" which tunnel all traffic through a local
VPN. These come also whith an understandable GUI.

My problem is, that these totally block traffic which is
initiated from outside the android device (which is a
good idea regarding firewalls).

I sideload all my programs (no contact to google
services for privacy reasons) by starting a webserver on
my tablet and send the data from my PC to the tablet --
which is now impossible because I am using a firewall.

My question is:
I only want to allow my tablet to connect via wifi
to my PC (I am using the "create_ap" script).
Due the firewall I need to fetch the data from
my PC instead of sending the data to my tablet.
It is sufficient to get access to only one directory
of my PC.
I dont want to setup a complete webserver on my PC
just for some Megs of data to transfer.

What is a secure, easy to setup and convenient method
to acchieve this?
(I only need some keywords ... no howtos needed. Will
try to setup the rest mysefl...;)

Thank you very much in advance for any help!

Best regards,
Meino

Re: [gentoo-user] Hard drive click of death?

[Daniel Frey]

On 10/03/2015 05:32 AM, Mick wrote:
> On Saturday 03 Oct 2015 12:23:47 Dale wrote:
>> Mick wrote:
>>> Hi All,
>>>
>>> Recently I noticed that an old drive is clicking 3 or 4 times when it is
>>> powered up.  Thereafter is stays quiet.  I think it is a mechanical
>>> fault, because I placed the box on its side and the drive did not click
>>> when booting up.
>>>
>>> I ran smartctl short/long/conveyance tests and no errors are reported.
>>>
>>> In your experience is this something to concern myself with?  There is
>>> not business critical data on the drive at present.

Some drives click from new, Toshiba laptop drives come to mind. However,
this Toshiba drive would only click when powered up. We are still using
this particular drive, and the laptop is 5-6 years old now. We tried
RMA'ing that drive and it came back no fault found, with a note that it
may click occasionally. (Design defect? Who knows...)

>>
>> I have found that if SMART reports a error, it is good to replace it as
>> soon as you can.  When a drive makes a noise that isn't normal, that's
>> also a sign that you need to replace it.  If you google around for that
>> model of drive, you may can find where others have had the same and they
>> shed some light on what happened, it died, it ran for ages and is normal
>> or something else.

I agree, check SMART regularly. Even in a crontab (use cronie if you
power off a lot so it will run if missed.) You can also set up email
monitoring so it will email when tests run.

>>
>> Right now, backups would be a good idea.  Doing some drive shopping
>> would to unless google turns up something that says it is nothing to
>> worry about, doubtful tho.

I have lost hard drives before and learned my lesson. My data is in
three different places, one being offline.

> 
> I don't know if I am getting more nostalgic in my old days, but it is things 
> like this that make me like spinning drives.  More often than not they give 
> some kind of warning.  :-)

I hate SSDs for this. I've had two outright fail with no warning. When
they failed even SMART reported no errors! But it was obvious the drives
weren't writing the data properly.

> 
> That said I've not yet had an SSD going sideways on me.
> 

You are lucky. I had one fail in my laptop, and one in my main mythtv
frontend. Two different brands too. :-(

Dan

Re: [gentoo-user] Hard drive click of death?

[Rich Freeman]

On Sat, Oct 3, 2015 at 12:36 PM, Daniel Frey  wrote:
>
> I have lost hard drives before and learned my lesson. My data is in
> three different places, one being offline.
>

++

All drives fail, and sooner or later all fingers fumble.  They may or
may not provide warning before it happens.

Don't backup periodically.  Test your backups periodically.  The
backups themselves should be automatic, very frequent, and at least
reasonably distant.

-- 
Rich

Re: [gentoo-user] Hard drive click of death?

[Meino . Cramer]

Mick  [15-10-03 11:44]:
> Hi All,
> 
> Recently I noticed that an old drive is clicking 3 or 4 times when it is 
> powered up.  Thereafter is stays quiet.  I think it is a mechanical fault, 
> because I placed the box on its side and the drive did not click when booting 
> up.
> 
> I ran smartctl short/long/conveyance tests and no errors are reported.
> 
> In your experience is this something to concern myself with?  There is not 
> business critical data on the drive at present.
> -- 
> Regards,
> Mick

Hi Mick,

I had a drive doing similiar things whith comparable "features".
In my case it was a defective cable, which gave good contact when
the drive was put on its side due the slight drill of the cable.
But it is less likely as what the other said, though...

HTH!
Best regards,
Meino

Re: [gentoo-user] Hard drive click of death?

[Mick]

On Saturday 03 Oct 2015 19:44:03 meino.cra...@gmx.de wrote:
> Mick  [15-10-03 11:44]:
> > Hi All,
> > 
> > Recently I noticed that an old drive is clicking 3 or 4 times when it is
> > powered up.  Thereafter is stays quiet.  I think it is a mechanical
> > fault, because I placed the box on its side and the drive did not click
> > when booting up.
> > 
> > I ran smartctl short/long/conveyance tests and no errors are reported.
> > 
> > In your experience is this something to concern myself with?  There is
> > not business critical data on the drive at present.
> 
> Hi Mick,
> 
> I had a drive doing similiar things whith comparable "features".
> In my case it was a defective cable, which gave good contact when
> the drive was put on its side due the slight drill of the cable.
> But it is less likely as what the other said, though...
> 
> HTH!
> Best regards,
> Meino

HA!  I reseated the cable on the drive and the MoBo and tried to start the PC 
a couple of times.  So far no clicking!  :-)

Let's hope it stays this way.

Thank you Meino for your suggestion.

-- 
Regards,
Mick


signature.asc
Description: This is a digitally signed message part.

Re: [gentoo-user] update problems

[allan gottlieb]

On Sat, Oct 03 2015, l...@yagibdah.de wrote:

> What is b.g.o.?

http://bugs.gentoo.org

allan

Re: [gentoo-user] Gentoo speaking to Android

[Bill Kenworthy]

On 04/10/15 02:04, meino.cra...@gmx.de wrote:
> Hi,
> 
> my setup is a recent Gentoo system, which is connected to the
> internet. This acts as an Wifi hotspot for my android tablet.

> 
> What is a secure, easy to setup and convenient method
> to acchieve this?
> (I only need some keywords ... no howtos needed. Will
> try to setup the rest mysefl...;)
> 
> Thank you very much in advance for any help!
> 
> Best regards,
> Meino
> 
> 
> 
> 
> 

I use an openvpn tunnel to bypass the firewall ... feed whatever you
like down the tunnel - I currently use a dav server and total commander
with its dav plugin.

Android and IOS are both security risks (from the point of you have no
control over what they send back to big brother and his numerous
friends) so isolate them from the rest of your network if necessary.

BillK

Re: [gentoo-user] Gentoo speaking to Android

[Meino . Cramer]

Bill Kenworthy  [15-10-04 04:08]:
> On 04/10/15 02:04, meino.cra...@gmx.de wrote:
> > Hi,
> > 
> > my setup is a recent Gentoo system, which is connected to the
> > internet. This acts as an Wifi hotspot for my android tablet.
> 
> > 
> > What is a secure, easy to setup and convenient method
> > to acchieve this?
> > (I only need some keywords ... no howtos needed. Will
> > try to setup the rest mysefl...;)
> > 
> > Thank you very much in advance for any help!
> > 
> > Best regards,
> > Meino
> > 
> > 
> > 
> > 
> > 
> 
> I use an openvpn tunnel to bypass the firewall ... feed whatever you
> like down the tunnel - I currently use a dav server and total commander
> with its dav plugin.
> 
> Android and IOS are both security risks (from the point of you have no
> control over what they send back to big brother and his numerous
> friends) so isolate them from the rest of your network if necessary.
> 
> BillK

Hi Bill,

(if you pull the word "openvpn" from ist context it looks like a
contradiction in itsself ;) ;))) 8)

Thank you very much for the informations! Yes, Android and others
looks like, as they were invented to spy on people. Fortunately,
Android can be rooted and there quite a few possibilties to find
out the apps, which are bad (at least).

I will see, how this fit all together. The Android firewall itsself
uses a device local VPN, so I dont know, whether this may clash with
the setup...I will see.

Thanks a lot! :)

Best regards,
Meino

Re: [gentoo-user] Question about initial/default ownership of /usr/portage

[Neil Bothwick]

On Fri, 2 Oct 2015 20:48:06 -0700, Raymond Jennings wrote:

> Who is supposed to own /usr/portage?

This was discussed in some detail two weeks ago. Search for the thread
entitled "portage directory ownerships?".


-- 
Neil Bothwick

All mail what i send is thoughly proof-red, definately!


pgpbobyNP7qmA.pgp
Description: OpenPGP digital signature