[gentoo-user] gcc 4.7.3 --> 4.8.3

[gentoo-user] gcc 4.7.3 --> 4.8.3

[James]

Ok 

so I'm still on 4.7.3; but if I set 4.8.3
as the default, should I rebuild @system ?

 # gcc-config -l
 [1] x86_64-pc-linux-gnu-4.7.3 *
 [2] x86_64-pc-linux-gnu-4.8.3


I saw the news item about 4.8.3-SSP, which I think is a good idea, but 
how deeply, if at all, do I need to rebuild packages ?
Is there any special steps I should take now in prepartion for 4.9.x?
caveats?

@system ?
@world  ?
a specific list  of packages only ?



James

Re: [gentoo-user] gcc 4.7.3 --> 4.8.3

[Mark Pariente]

> so I'm still on 4.7.3; but if I set 4.8.3
> as the default, should I rebuild @system ?
> 
>  # gcc-config -l
>  [1] x86_64-pc-linux-gnu-4.7.3 *
>  [2] x86_64-pc-linux-gnu-4.8.3
> 
> 
> I saw the news item about 4.8.3-SSP, which I think is a good idea, but
> how deeply, if at all, do I need to rebuild packages ?
> Is there any special steps I should take now in prepartion for 4.9.x?
> caveats?
> 
> @system ?
> @world  ?
> a specific list  of packages only ?

Based on my experience going from 4.7 to 4.8 was seamless, you can 
choose to not rebuild anything if you'd like. If you want to start 
taking advantage of -fstack-protector by default you can rebuild 
whatever you'd like, although if you're paranoid about security you 
should rebuild everything, ie. @system @world to make sure all your 
binaries and the libraries they dynamically load have the stack 
smashing guard built in.

Going to 4.9 though is another thing. Apparently they broke the ABI for 
the standard C++ library, so once you start compiling C++ stuff with 
4.9 you better go all in (I did @system @world with 4.9 and had very 
few things that failed to compile[1], it's looking pretty good already).

--Mark

[1] https://bugs.gentoo.org/show_bug.cgi?id=526140

Re: [gentoo-user] gcc 4.7.3 --> 4.8.3

[Todd Goodman]

* James <wireless@tampabay.rr.com> [141107 12:47]:
> Ok 
> 
> so I'm still on 4.7.3; but if I set 4.8.3
> as the default, should I rebuild @system ?
> 
>  # gcc-config -l
>  [1] x86_64-pc-linux-gnu-4.7.3 *
>  [2] x86_64-pc-linux-gnu-4.8.3
> 
> 
> I saw the news item about 4.8.3-SSP, which I think is a good idea, but 
> how deeply, if at all, do I need to rebuild packages ?
> Is there any special steps I should take now in prepartion for 4.9.x?
> caveats?
> 
> @system ?
> @world  ?
> a specific list  of packages only ?
> 
> 
> 
> James

If I remember correctly (and I may not) the only thing I *had* to
rebuild was webkit-gtk with 4.8 as I was getting errors with apps built
with 4.8 and webkit-gtk built by 4.7.

But I tend to set up a @world rebuild to run when I'm not on my machines
if I have any issues anyway so I may not have run into other problems I
might have had otherwise.

Todd

[gentoo-user] Re: gcc 4.7.3 --> 4.8.3

[Nikos Chantziaras]

On 07/11/14 19:46, James wrote:
> Ok
>
> so I'm still on 4.7.3; but if I set 4.8.3
> as the default, should I rebuild @system ?
>
>   # gcc-config -l
>   [1] x86_64-pc-linux-gnu-4.7.3 *
>   [2] x86_64-pc-linux-gnu-4.8.3
>
>
> I saw the news item about 4.8.3-SSP, which I think is a good idea, but
> how deeply, if at all, do I need to rebuild packages ?

You don't need to rebuild, although there are known problems with having 
both 4.7 and 4.8 installed, and having 4.7 be the active one (or any 
case where an older version is the active one.)

You should be able to just switch to 4.8 without rebuilding anything. 
That's what I did. Of course it can't hurt to rebuild everything, but 
you can schedule that for later (like an overnight rebuild of @world 
with --keep-going). It's not critical to do it immediately.

Re: [gentoo-user] gcc 4.7.3 --> 4.8.3

[Stefan G. Weichinger]

Am 07.11.2014 um 19:19 schrieb Mark Pariente:

> Going to 4.9 though is another thing. Apparently they broke the ABI for
> the standard C++ library, so once you start compiling C++ stuff with 4.9
> you better go all in (I did @system @world with 4.9 and had very few
> things that failed to compile[1], it's looking pretty good already).
> 
> --Mark
> 
> [1] https://bugs.gentoo.org/show_bug.cgi?id=526140

I give that a try ... as I set up a fresh new btrfs-subvolume to do a
fresh new build based on my @world only yesterday I will see if I can do
it with gcc 4.9 while I am at it.

Let's see if things work out and if it gets any better ;-)

[gentoo-user] Re: gcc 4.7.3 --> 4.8.3

[James]

Nikos Chantziaras <realnc <at> gmail.com> writes:


> > I saw the news item about 4.8.3-SSP, which I think is a good idea, but
> > how deeply, if at all, do I need to rebuild packages ?

> You don't need to rebuild, although there are known problems with having 
> both 4.7 and 4.8 installed, and having 4.7 be the active one (or any 
> case where an older version is the active one.)

Hmmmmm.

> You should be able to just switch to 4.8 without rebuilding anything. 
> That's what I did. Of course it can't hurt to rebuild everything, but 
> you can schedule that for later (like an overnight rebuild of  <at> world 
> with --keep-going). It's not critical to do it immediately.

OK, so I use updated (sync/compile) using 4.7.3.  I did have to
rebuild webkit-gtk, as todd suggested. Now I'm going to switch
over to 4.8.3 and rebuild @sytem and @world.  These are fun
and not big of a deal with my FX8350 and 32g or ram.....

thx for the advise, everyone,


James

Re: [gentoo-user] gcc 4.7.3 --> 4.8.3

[Volker Armin Hemmann]

Am 07.11.2014 um 19:19 schrieb Mark Pariente:
>
>
> Going to 4.9 though is another thing. Apparently they broke the ABI
> for the standard C++ library, so once you start compiling C++ stuff
> with 4.9 you better go all in (I did @system @world with 4.9 and had
> very few things that failed to compile[1], it's looking pretty good
> already).

oh great, again? Is it that time of the month?

Re: [gentoo-user] Re: gcc 4.7.3 --> 4.8.3

[Peter Humphrey]

On Saturday 08 November 2014 18:17:02 Nikos Chantziaras wrote:
> On 07/11/14 19:46, James wrote:
> > Ok
> > 
> > so I'm still on 4.7.3; but if I set 4.8.3
> > as the default, should I rebuild @system ?
> > 
> >   # gcc-config -l
> >   [1] x86_64-pc-linux-gnu-4.7.3 *
> >   [2] x86_64-pc-linux-gnu-4.8.3
> > 
> > I saw the news item about 4.8.3-SSP, which I think is a good idea, but
> > how deeply, if at all, do I need to rebuild packages ?
> 
> You don't need to rebuild, although there are known problems with having
> both 4.7 and 4.8 installed, and having 4.7 be the active one (or any
> case where an older version is the active one.)
> 
> You should be able to just switch to 4.8 without rebuilding anything.
> That's what I did. Of course it can't hurt to rebuild everything, but
> you can schedule that for later (like an overnight rebuild of @world
> with --keep-going). It's not critical to do it immediately.

I'd have thought you needed to emerge -e world if you really want to be 
protected.

-- 
Rgds
Peter.

[gentoo-user] Re: gcc 4.7.3 --> 4.8.3

[James]

Peter Humphrey <peter <at> prh.myzen.co.uk> writes:


> > You should be able to just switch to 4.8 without rebuilding anything.
> > That's what I did. Of course it can't hurt to rebuild everything, but
> > you can schedule that for later (like an overnight rebuild of  <at> world
> > with --keep-going). It's not critical to do it immediately.

That's exactly what I did, more or less. All seems fine.

> I'd have thought you needed to emerge -e world if you really want to be 
> protected.

Yea, maybe. I read the man page on emptytree. I get it actually replaces
by a "reinstall".   Does this do more than if I just reboot after

emerge @system @world and then reboot?  

I'd be curious to know exactly what reinstall does that is not covered
by just starting up a given code again? 

Is it that it forces a reinstall and stop/starts the binary without
rebooting?   

Rebooting catches *everything* even better than --emptytree ?


James

Re: [gentoo-user] Re: gcc 4.7.3 --> 4.8.3

[Neil Bothwick]

[-- Attachment #1: Type: text/plain, Size: 1163 bytes --]

On Mon, 10 Nov 2014 18:52:09 +0000 (UTC), James wrote:

> > I'd have thought you needed to emerge -e world if you really want to
> > be protected.  
> 
> Yea, maybe. I read the man page on emptytree. I get it actually replaces
> by a "reinstall".   Does this do more than if I just reboot after
> 
> emerge @system @world and then reboot?  
> 
> I'd be curious to know exactly what reinstall does that is not
> covered by just starting up a given code again? 
> 
> Is it that it forces a reinstall and stop/starts the binary without
> rebooting?   
> 
> Rebooting catches *everything* even better than --emptytree ?

--emptytree has nothing to do with rebooting. It simply forces emerge to
rebuild everything in @world and their dependencies. Once you have done
that, you will have daemons still running the old code, which you could
fix with a reboot, or you could run checkrestart and restart only the
affected programs.

After an emerge -e @world, a reboot is probably best, another reason to
avoid the unnecessary step of emerge -e @world in the first place.


-- 
Neil Bothwick

Top Oxymorons Number 20: Synthetic natural gas

[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 181 bytes --]

Re: [gentoo-user] Re: gcc 4.7.3 --> 4.8.3

[Tomas Mozes]

On 2014-11-10 23:23, Neil Bothwick wrote:
> On Mon, 10 Nov 2014 18:52:09 +0000 (UTC), James wrote:
> 
>> > I'd have thought you needed to emerge -e world if you really want to
>> > be protected.
>> 
>> Yea, maybe. I read the man page on emptytree. I get it actually 
>> replaces
>> by a "reinstall".   Does this do more than if I just reboot after
>> 
>> emerge @system @world and then reboot?
>> 
>> I'd be curious to know exactly what reinstall does that is not
>> covered by just starting up a given code again?
>> 
>> Is it that it forces a reinstall and stop/starts the binary without
>> rebooting?
>> 
>> Rebooting catches *everything* even better than --emptytree ?
> 
> --emptytree has nothing to do with rebooting. It simply forces emerge 
> to
> rebuild everything in @world and their dependencies. Once you have done
> that, you will have daemons still running the old code, which you could
> fix with a reboot, or you could run checkrestart and restart only the
> affected programs.
> 
> After an emerge -e @world, a reboot is probably best, another reason to
> avoid the unnecessary step of emerge -e @world in the first place.

Or you can check the list of processes with deleted libraries and 
restart those.

lsof -n | grep 'DEL.*lib'

Re: [gentoo-user] Re: gcc 4.7.3 --> 4.8.3

[Dale]

Neil Bothwick wrote:
> On Mon, 10 Nov 2014 18:52:09 +0000 (UTC), James wrote:
>
>>> I'd have thought you needed to emerge -e world if you really want to
>>> be protected.  
>> Yea, maybe. I read the man page on emptytree. I get it actually replaces
>> by a "reinstall".   Does this do more than if I just reboot after
>>
>> emerge @system @world and then reboot?  
>>
>> I'd be curious to know exactly what reinstall does that is not
>> covered by just starting up a given code again? 
>>
>> Is it that it forces a reinstall and stop/starts the binary without
>> rebooting?   
>>
>> Rebooting catches *everything* even better than --emptytree ?
> --emptytree has nothing to do with rebooting. It simply forces emerge to
> rebuild everything in @world and their dependencies. Once you have done
> that, you will have daemons still running the old code, which you could
> fix with a reboot, or you could run checkrestart and restart only the
> affected programs.
>
> After an emerge -e @world, a reboot is probably best, another reason to
> avoid the unnecessary step of emerge -e @world in the first place.
>
>

After I do a major upgrade or --emptytree, I switch to boot runlevel,
check with checkrestart and restart whatever it reports needs it. 
Generally, switching to boot runlevel catches most everything.

Yea, rebooting may be faster but I hate rebooting all the time.  :/

Dale

:-)  :-)

[gentoo-user] Re: gcc 4.7.3 --> 4.8.3

[James]

Dale <rdalek1967 <at> gmail.com> writes:

> 
> Neil Bothwick wrote:
> >> Rebooting catches *everything* even better than --emptytree ?
> > --emptytree has nothing to do with rebooting. It simply forces emerge to
> > rebuild everything in  <at> world and their dependencies. Once you have
> > done, you will have daemons still running the old code, which you could
> > fix with a reboot, or you could run checkrestart and restart only the
> > affected programs.

Ah, that is what I thought.

> > After an emerge -e  <at> world, a reboot is probably best, another 
> > reason to avoid the unnecessary step of emerge -e  <at> world in the 
> > first place.

This conflict what others have said. Curious. My take is that since
I updated the major compiler, gcc, it warrants an --emptytree rebuild
and reboot, just to be safe. It's a workstation, not a server, so it's
time for a reboot, imho.

> After I do a major upgrade or --emptytree, I switch to boot runlevel,
> check with checkrestart and restart whatever it reports needs it. 
> Generally, switching to boot runlevel catches most everything.

OK, so I emerge checkrestart and ran it. And there are almost a dozen things
it says need a reboot (mostly lxde). "These processes do not seem to have an
associated init script to restart them".

So I have to reboot anyways. Oh, the url on the "checkrestart" script
now points to some advertisement that is unrelated, to a bug needs to
be file to the github location? I did not know if this is the best new
link, so I did not file this bug on checkrestart.



*******************
> Yea, rebooting may be faster but I hate rebooting all the time.  :/

Agreeded. But after a gcc update, I think it wise, especially since
gcc-4.9 cometh....soon?

> Dale

thx,
James

Re: [gentoo-user] Re: gcc 4.7.3 --> 4.8.3

[Neil Bothwick]

[-- Attachment #1: Type: text/plain, Size: 2032 bytes --]

On Tue, 11 Nov 2014 20:19:36 +0000 (UTC), James wrote:

> Dale <rdalek1967 <at> gmail.com> writes:
> > Neil Bothwick wrote:

> > > After an emerge -e  <at> world, a reboot is probably best, another 
> > > reason to avoid the unnecessary step of emerge -e  <at> world in
> > > the first place.
> 
> This conflict what others have said. Curious. My take is that since
> I updated the major compiler, gcc, it warrants an --emptytree rebuild
> and reboot, just to be safe.

Why? The compiler is not used by running software. If there was an ABI
change meaning that mixing programs compiled with the two versions would
cause problem, emerge -e would be prudent, but that hasn't happened for a
long time. You don't dismantle and reassenble your car just because you
bought a new set of spanners...

> > After I do a major upgrade or --emptytree, I switch to boot runlevel,
> > check with checkrestart and restart whatever it reports needs it. 
> > Generally, switching to boot runlevel catches most everything.
> 
> OK, so I emerge checkrestart and ran it. And there are almost a dozen
> things it says need a reboot (mostly lxde). "These processes do not
> seem to have an associated init script to restart them".
> 
> So I have to reboot anyways.

No, simply log out of the desktop and back in.

Bear in mind that some of what checkrestart reports is unimportant
anyway. Just because a process is using a slightly older in-memory
version of a library doesn't mean it is suddenly going to stop working. I
have services that have been flagged by checkrestart for weeks that are
still fine, I just don't want to stop and restart them.

> > Yea, rebooting may be faster but I hate rebooting all the time.  :/
> 
> Agreeded. But after a gcc update, I think it wise, especially since
> gcc-4.9 cometh....soon?

Yes, things may be a little different with 4.9, but the last time a
rebuild was really required was,AFAIR, somewhere around 3.3.


-- 
Neil Bothwick

Dyslexics of the world, untie!

[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 181 bytes --]

[gentoo-user] Re: gcc 4.7.3 --> 4.8.3

[James]

Tomas Mozes <tomas.mozes <at> shmu.sk> writes:


> >> Rebooting catches *everything* even better than --emptytree ?

> > After an emerge -e  <at> world, a reboot is probably best, another  
  > > reason to
> > avoid the unnecessary step of emerge -e  <at> world in the first place.

> Or you can check the list of processes with deleted libraries and 
> restart those. 

> lsof -n | grep 'DEL.*lib'


Nice use of lsof. However it reveals way too much stuff to deal
with manually. The reboot will be much faster. Just so you know,
I've been hacking java, scala and such and my /usr/local/portage
is growing everyday. So a clean reboot is most warranted......

after an --emptytree    later on tonight..... 
(thank god for a FX8350 and 32 gig of ram........)


Thanks for all the good advice,
James

Re: [gentoo-user] Re: gcc 4.7.3 --> 4.8.3

[Mick]

[-- Attachment #1: Type: Text/Plain, Size: 705 bytes --]

On Tuesday 11 Nov 2014 21:03:56 Neil Bothwick wrote:

> Why? The compiler is not used by running software. If there was an ABI
> change meaning that mixing programs compiled with the two versions would
> cause problem, emerge -e would be prudent, but that hasn't happened for a
> long time. You don't dismantle and reassenble your car just because you
> bought a new set of spanners...

Hmmm ...  O_o

I can't recall when I rebuilt system (or was it world) because gcc changed, 
but if I were to guess I would say it was back in 2004/5.  I don't recall 
experiencing problems waiting for packages to rebuild themselves when a new 
version became available in portage.

-- 
Regards,
Mick

[-- Attachment #2: This is a digitally signed message part. --]
[-- Type: application/pgp-signature, Size: 473 bytes --]

[gentoo-user] Re: gcc 4.7.3 --> 4.8.3

[James]

> > > Neil Bothwick wrote:

> > > > After an emerge -e  <at> world, a reboot is probably best, another 
> > > > reason to avoid the unnecessary step of emerge -e  <at> world in
> > > > the first place.
  


> > This conflict what others have said. Curious. My take is that since
> > I updated the major compiler, gcc, it warrants an --emptytree rebuild
> > and reboot, just to be safe.

> Why? The compiler is not used by running software. If there was an ABI
> change meaning that mixing programs compiled with the two versions would
> cause problem, emerge -e would be prudent, but that hasn't happened for a
> long time. You don't dismantle and reassenble your car just because you
> bought a new set of spanners...
> 

Re: [gentoo-dev] more help needed with gcc-4.8 stabilization

It's a long thread, and not the ony one that hints at issues
of installing 4.8.x and still having 4.7.3 set as the default.
"webkit-gtk" was one that took me a few tries to get to compile
completely. ymmv.

Obviously many things have been resolved that are listed in the thread.
" have you considered to stabilize gcc:4.9 instead possibly 4.9.2 ?
I'm not really suggesting to do so, but seem that most of the problems
of 4.9.1 are the same of 4.8.3 so maybe it's worth considering. "

that said 4.8.3 is marked stable by the devs, but a large part of that
is 4.9.x is needed by some "key" codes coming down the pipe. No, I did
not write thus down, just made myself a mental note up upgrade everything
to 4.8.3 in preparation for 4.9.x. (Chrome is on, I think).

> > > After I do a major upgrade or --emptytree, I switch to boot runlevel,
> > > check with checkrestart and restart whatever it reports needs it. 
> > > Generally, switching to boot runlevel catches most everything.
> > 
> > OK, so I emerge checkrestart and ran it. And there are almost a dozen
> > things it says need a reboot (mostly lxde). "These processes do not
> > seem to have an associated init script to restart them".
> > 
> > So I have to reboot anyways.
> 
> No, simply log out of the desktop and back in.

Um, Tomas's little one-liner:
lsof -n | grep 'DEL.*lib'

revealed far to much to deal with. I got lib issues coming out of my arse
(I've been hacking at a few things I do not fully understand 
(wink wink :: nudge nudge) ?


> Bear in mind that some of what checkrestart reports is unimportant
> anyway. Just because a process is using a slightly older in-memory
> version of a library doesn't mean it is suddenly going to stop working. I
> have services that have been flagged by checkrestart for weeks that are
> still fine, I just don't want to stop and restart them.

Granted. My need to reboot is because I've been noodling around with
many many things. My current desktop: lxde is crippled and deprecated.
Lx1t-0.8.0 is in the tree now, but masked waiting on another package 
or 2 to be tweaked.



> Yes, things may be a little different with 4.9, but the last time a
> rebuild was really required was,AFAIR, somewhere around 3.3.


OK, so I reboot workstations more often than you. I hope that does
not upset you? Yes, I've kept workstations online for over a year more
times that I can count (fingers and toes). And when the reboot comes, It's a
day or 2 fixing things, imho. YMMV. A judicious reboot now and again, timed
well, is keenly a good idea, imho. ymmv. Besides I'm an old FT via
redundancy, kind of guy; aka I *always* have spare systems, ready to go.


On the server side. When I have to be "responsible" for servers others
use, I *always* have duplicated hot spares, or I don't do it. I'm not
saying that other should/have to do what I do. I'm very lazy and only
get lucky when it counts. No I'm too forgetful to be considered smart
anymore. So, I use spare hardwares, boot them up and away I go! I live
in Florida; so the power failures can "jump" UPS's, ethernet cables and
all sorts of strange issues, not just admin issues dictace FT via
redundancy for me. I may just move my shop onto a sail boat, so then
I'd have metal_chloride issuse to deal with......


So, via hardware redudancy, as thecomplete system level,  I can 
diagnose failures at my liesure, sipping coffee, wine
or a beverage that would make Alan crazy (quite a few of these....).

I deeply appreciate your concerns over the admin skills of an old_fart....
RFC 5798. 

cheers?
James

Re: [gentoo-user] Re: gcc 4.7.3 --> 4.8.3

[Dale]

James wrote:
> Dale <rdalek1967 <at> gmail.com> writes:
>
>> After I do a major upgrade or --emptytree, I switch to boot runlevel,
>> check with checkrestart and restart whatever it reports needs it. 
>> Generally, switching to boot runlevel catches most everything.
> OK, so I emerge checkrestart and ran it. And there are almost a dozen things
> it says need a reboot (mostly lxde). "These processes do not seem to have an
> associated init script to restart them".
>
> So I have to reboot anyways. Oh, the url on the "checkrestart" script
> now points to some advertisement that is unrelated, to a bug needs to
> be file to the github location? I did not know if this is the best new
> link, so I did not file this bug on checkrestart.
>
>
>
> *******************
>> Yea, rebooting may be faster but I hate rebooting all the time.  :/
> Agreeded. But after a gcc update, I think it wise, especially since
> gcc-4.9 cometh....soon?
>
>> Dale
> thx,
> James
>
>

If checkrestart says there is no init scripts to restart the process,
odds are you don't need to really worry about it.  That said, I use htop
to find out what is running and do what I can to restart them anyway. 
Usually when I get that message, restarting udev or lvmetad gives me a
clean output. 

Usually if a gcc is released that requires all this, it is well known. 
Any gotchas related to gcc will spread like fire.  There is to many
people using gcc for something of that nature to sneak by plus the
coders usually know this and make it known as well.   It's been a good
while since the rebuilding of everything has been required.  I tend to
do it myself but I don't get my hands to dirty over the deal.  I usually
do it the next time there is a KDE upgrade or something since that
rebuilds a lot of packages anyway. 

Just something to ponder on.  ;-)

Dale

:-)  :-) 

Re: [gentoo-user] Re: gcc 4.7.3 --> 4.8.3

[Neil Bothwick]

[-- Attachment #1: Type: text/plain, Size: 1889 bytes --]

On Wed, 12 Nov 2014 02:07:23 +0000 (UTC), James wrote:

> > No, simply log out of the desktop and back in.  
> 
> Um, Tomas's little one-liner:
> lsof -n | grep 'DEL.*lib'
> 
> revealed far to much to deal with. I got lib issues coming out of my
> arse (I've been hacking at a few things I do not fully understand 
> (wink wink :: nudge nudge) ?

Oh yes, there are times when a reboot is the simplest option, it makes
sure everything is back as baseline. I have no problem with rebooting,
just wouldn't recommend the worldwide recompile that leads to the need to
do so - unless it is necessary, which is clearly is not in the case of
the GCC 4.7 -> 4.8 switch. I run ~arch on my desktop, so most things will
be recompiled in the next few weeks anyway.

> > Yes, things may be a little different with 4.9, but the last time a
> > rebuild was really required was,AFAIR, somewhere around 3.3.  
 
> OK, so I reboot workstations more often than you. I hope that does
> not upset you?

As long as they are your workstations and not mine,m it doesn't upset me
in the least ;-)

> Yes, I've kept workstations online for over a year more
> times that I can count (fingers and toes). And when the reboot comes,
> It's a day or 2 fixing things, imho. YMMV.

Not by much. Severs may stay up for a while but desktops tend to be
rebooted quite often after kernel updates.

> A judicious reboot now and
> again, timed well, is keenly a good idea, imho. ymmv. Besides I'm an
> old FT via redundancy, kind of guy; aka I *always* have spare systems,
> ready to go.

As a fully paid up member of the Old Farts Club, I can appreciate that.
 
> I deeply appreciate your concerns over the admin skills of an
> old_fart.... RFC 5798. 

I don't see how that RFC relates to being an old fart...


-- 
Neil Bothwick

I backed up my hard drive and ran into a bus.

[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 181 bytes --]

[gentoo-user] Re: gcc 4.7.3 --> 4.8.3

[Jonathan Callen]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA512

On 11/11/2014 04:03 PM, Neil Bothwick wrote:
> On Tue, 11 Nov 2014 20:19:36 +0000 (UTC), James wrote:
>> 
>> Agreeded. But after a gcc update, I think it wise, especially 
>> since gcc-4.9 cometh....soon?
> 
> Yes, things may be a little different with 4.9, but the last time a
> rebuild was really required was,AFAIR, somewhere around 3.3.
> 
> 

The last time a rebuild of (almost) everything was required was when
the C++ ABI changed, with the associated bump of SONAME from
libstdc++.so.5 (provided with GCC 3.3 and earlier) to libstdc++.so.6
(provided with GCC 3.4 and later).  So you were close, but the major
change happened with 3.4, not 3.3 ;).

Some old binary software still requires libstdc++.so.5, which can
still be installed from sys-libs/libstdc++-v3, which actually builds
part of GCC 3.3.6 to get the libstdc++.so.5 to install.

- -- 
Jonathan Callen
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2
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=5KgF
-----END PGP SIGNATURE-----

Re: [gentoo-user] Re: gcc 4.7.3 --> 4.8.3

[Neil Bothwick]

[-- Attachment #1: Type: text/plain, Size: 644 bytes --]

On Thu, 13 Nov 2014 23:52:22 -0500, Jonathan Callen wrote:

> > Yes, things may be a little different with 4.9, but the last time a
> > rebuild was really required was,AFAIR, somewhere around 3.3.

> The last time a rebuild of (almost) everything was required was when
> the C++ ABI changed, with the associated bump of SONAME from
> libstdc++.so.5 (provided with GCC 3.3 and earlier) to libstdc++.so.6
> (provided with GCC 3.4 and later).  So you were close, but the major
> change happened with 3.4, not 3.3 ;).

Obviously, I meant FROM 3.3 :P

;-)


-- 
Neil Bothwick

I'm writing a book. I've got the page numbers done.

[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 181 bytes --]