11.0-RC1 unsupported by ports?

Daniel Braniss danny at cs.huji.ac.il
Wed Jan 25 11:28:09 UTC 2017


> On 25 Jan 2017, at 12:39, Bernd Walter <ticso at cicely7.cicely.de> wrote:
> 
> On Wed, Jan 25, 2017 at 11:52:10AM +0200, Daniel Braniss wrote:
>> 
>>> On 25 Jan 2017, at 10:47, Bernd Walter <ticso at cicely7.cicely.de> wrote:
>>> 
>>> On Wed, Jan 25, 2017 at 09:13:18AM +0100, Kurt Jaeger wrote:
>>>> Hi!
>>>> 
>>>>>> 11.0-RC1 was superseded by 11.0-REL, so while that message is a bit
>>>>>> drastic, there's a point to it.
>>>>> 
>>>>> With that argument only the latest version would be supported.
>>>> 
>>>> https://www.freebsd.org/releases/ lists the supported releases.
>>>> There are no release candidates listed.
>>>> 
>>>>> That said, it is a release candidate and as such one could argue that
>>>>> there never had been any official support at all.
>>>>> In that case however the message is wrong, because when a support has
>>>>> ended it implies that there was support.
>>>>> 
>>>>> The check in the code is this one:
>>>>> .if (${OPSYS} == FreeBSD && (${OSVERSION} < 1003000 || (${OSVERSION} >= 1100000 && ${OSVERSION} < 1100122))) || \
>>>>>   (${OPSYS} == DragonFly && ${DFLYVERSION} < 400400)
>>>>> 
>>>>> It is not about RC as such, it is explicitly about 11.0-RC.
>>>>> My OSVERSION is 1100121.
>>>>> So obviously support starts with the first release.
>>>>> Fair enough, but then the message is still wrong unless it was supported.
>>>> 
>>>> What's stopping you from upgrading to -REL ?
>>> 
>>> Buildworld on a raspberry isn't fun - if it works at all.
>>> Even if you crossbuild and just copy the binaries, the wear of
>>> MicroSD cards isn't something you want to test unless you really
>>> have to.
>> 
>> most of the time this works for me:
>> 	mount host:/export-to-rpi/local /usr/local
>> 	echo ???WRKDIRPREFIX=/var/tmp??? >> /etc/make.conf
>> 	mount via nfs /var/tmp, i.e.
>> 		mount host:/export-to-rpi/tmp  /var/tmp
>> 	also add swap via nfs:
>> 		mount host:/export-to-rpi/swap /mnt-swap
>> 		swapon /mnt-swap
> 
> This has nothing to do with updating the OS itself.
somehow the subject was ‘compiling ports’ to build the os I cross compile
and that works fine.

> 
> That said, I assume host:/export-to-rpi/local is only used by a
> single host.
no, if you take care, we have been sharing /usr/local since way back.
btw, it’s read only for the clients.

> It gets tricky with shared /usr/local, since the package registration
> is in a different path and ports/packages may also touch /etc - e.g.
> /etc/shells, or add service users for a specific software.
> It is possible to do, but unless you are very carefull things can
> easily get messy.
the ports compilation keep things in /var/db, and that tends to cause problems
with pkg - since it insists in locking the sqlite db, so I see that on the SD, and copy it
to more secure location when I’m done (I keep losing track of those small sd cards,
there is hardly any place to put a label on them :-)


> Same goes for /tmp.
> Needless to say that swap isn't to be shared at all…
that would certainly be foolish :-)

> But I'm not sure if swap on NFS is completely deadlock free.

so far It’s been just fine, e.g., compiling pkg takes up a lot of resources
and this was the only way it worked, not to mention gcc. I compiled all
of the X11! 

> 
> -- 
> B.Walter <bernd at bwct.de> http://www.bwct.de
> Modbus/TCP Ethernet I/O Baugruppen, ARM basierte FreeBSD Rechner uvm.



More information about the freebsd-arm mailing list