[ANN] Squeak 5.1 Code Freeze -- Trunk still closed; first release candidate(s) available

Previous Topic Next Topic
 
classic Classic list List threaded Threaded
18 messages Options
Reply | Threaded
Open this post in threaded view
|

[ANN] Squeak 5.1 Code Freeze -- Trunk still closed; first release candidate(s) available

marcel.taeumel
Hi, there.

It's time for another "code freeze". :-) Let's hope that all serious bugs got fixed by now and that we have a usable Squeak 5.1 environment to get creative with. ;-)

Find the first release candidate here:
http://files.squeak.org/5.1rc1/Squeak5.1rc1-16520-32bit/
http://files.squeak.org/5.1rc1/Squeak5.1rc1-16520-64bit/ (regarded as "experimental")

You can still tell us about every little "buglette" you can find. We will, however, address only "show stoppers" in this phase. Other things will be scribbled down in the release notes under "Known Issues" and (maybe) fixed after the release:
https://github.com/squeak-smalltalk/squeak-app/tree/master/release-notes

Please stress-test this RC1 until August 21, 23:59 AOE. Plus/minus a several hours as usual. :-D

Best,
Marcel
Reply | Threaded
Open this post in threaded view
|

Re: [ANN] Squeak 5.1 Code Freeze -- Trunk still closed; first release candidate(s) available

Bert Freudenberg
On Thu, Aug 18, 2016 at 9:02 AM, marcel.taeumel <[hidden email]> wrote:
Hi, there.

It's time for another "code freeze". :-) Let's hope that all serious bugs
got fixed by now and that we have a usable Squeak 5.1 environment to get
creative with. ;-)

Find the first release candidate here:
http://files.squeak.org/5.1rc1/Squeak5.1rc1-16520-32bit/
http://files.squeak.org/5.1rc1/Squeak5.1rc1-16520-64bit/ (regarded as
"experimental")

You can also try it in SqueakJS simply by clicking here:


- Bert -


Reply | Threaded
Open this post in threaded view
|

Re: [ANN] Squeak 5.1 Code Freeze -- Trunk still closed; first release candidate(s) available

timrowledge
In reply to this post by marcel.taeumel

> On 18-08-2016, at 12:02 AM, marcel.taeumel <[hidden email]> wrote:
>
> Hi, there.
>
> It's time for another "code freeze". :-) Let's hope that all serious bugs
> got fixed by now and that we have a usable Squeak 5.1 environment to get
> creative with. ;-)
>
> Find the first release candidate here:
> http://files.squeak.org/5.1rc1/Squeak5.1rc1-16520-32bit/

In the ARM version the shell script has problems that stop it working. Basically the regexp phrase isn’t doing quite what we need. See for example the log of what it did -
++ uname -r
+ local kernel_release=4.4.9-v7+
+ local 're=[^0-9]*\([0-9]*\)[.]\([0-9]*\)[.]\([0-9]*\)\([0-9A-Za-z-]*\)'
++ echo 4.4.9-v7+
++ sed -e 's#[^0-9]*\([0-9]*\)[.]\([0-9]*\)[.]\([0-9]*\)\([0-9A-Za-z-]*\)#\1#'
+ local major=4+
++ echo 4.4.9-v7+
++ sed -e 's#[^0-9]*\([0-9]*\)[.]\([0-9]*\)[.]\([0-9]*\)\([0-9A-Za-z-]*\)#\2#'
+ local minor=4+
++ echo 4.4.9-v7+
++ sed -e 's#[^0-9]*\([0-9]*\)[.]\([0-9]*\)[.]\([0-9]*\)\([0-9A-Za-z-]*\)#\3#'
+ local patch=9+
+ local min_major=2
+ local min_minor=6
+ local min_patch=12
+ [[ 4+ -lt 2 ]]
Squeak5.1rc1-16520-32bit-201608171728-ARMv6/squeak.sh: line 38: [[: 4+: syntax error: operand expected (error token is "+")
+ [[ 4+ -le 2 ]]
Squeak5.1rc1-16520-32bit-201608171728-ARMv6/squeak.sh: line 39: [[: 4+: syntax error: operand expected (error token is "+")
+ [[ 4+ -le 2 ]]

Note how the various major/minor variables are 4+ and not 4 etc. I guess this is what is causing the comparisons to fail? That ‘re’ needs some massaging.

tim
--
tim Rowledge; [hidden email]; http://www.rowledge.org/tim
For every action, there is an equal and opposite criticism.



Reply | Threaded
Open this post in threaded view
|

Re: [ANN] Squeak 5.1 Code Freeze -- Trunk still closed; first release candidate(s) available

timrowledge

> On 18-08-2016, at 5:13 PM, tim Rowledge <[hidden email]> wrote:
> {snip}
>
> Note how the various major/minor variables are 4+ and not 4 etc. I guess this is what is causing the comparisons to fail? That ‘re’ needs some massaging.

To get past that quickly I commented out the broken bits. It then proceeded to try to create the security/limits file even though it isn’t required. The ensure_conf_file needs to depend upon the kernel level.
I had a secondary problem due to having unzipped the package on my Mac which left .AppleDouble files around; ideally that ought not cause a problem if the `find` call is a bit cleverer. Why does it descend into .AppleDouble before checking the local dir?


tim
--
tim Rowledge; [hidden email]; http://www.rowledge.org/tim
Oxymorons: Good grief



Reply | Threaded
Open this post in threaded view
|

Re: [ANN] Squeak 5.1 Code Freeze -- Trunk still closed; first release candidate(s) available

fniephaus
On Fri, Aug 19, 2016 at 2:34 AM tim Rowledge <[hidden email]> wrote:

> On 18-08-2016, at 5:13 PM, tim Rowledge <[hidden email]> wrote:
> {snip}
>
> Note how the various major/minor variables are 4+ and not 4 etc. I guess this is what is causing the comparisons to fail? That ‘re’ needs some massaging.

To get past that quickly I commented out the broken bits. It then proceeded to try to create the security/limits file even though it isn’t required. The ensure_conf_file needs to depend upon the kernel level.
I had a secondary problem due to having unzipped the package on my Mac which left .AppleDouble files around; ideally that ought not cause a problem if the `find` call is a bit cleverer. Why does it descend into .AppleDouble before checking the local dir?

Thanks for the bug reports. I've fixed the regex bug and found a way to ensure "find" ignores dot files.
But why do you think ensure_conf_file needs to depend upon the kernel level? Would it be better to skip ensure_conf_file if we are on ARM? And why don't we need such a file on a Pi with kernel version 4+?

Fabio
 


tim
--
tim Rowledge; [hidden email]; http://www.rowledge.org/tim
Oxymorons: Good grief





Reply | Threaded
Open this post in threaded view
|

Re: [ANN] Squeak 5.1 Code Freeze -- Trunk still closed; first release candidate(s) available

timrowledge

> On 19-08-2016, at 12:41 PM, Fabio Niephaus <[hidden email]> wrote:
>
> Thanks for the bug reports. I've fixed the regex bug and found a way to ensure "find" ignores dot files.

Excellent.

> But why do you think ensure_conf_file needs to depend upon the kernel level?

Because the issue was fixed in the kernel some time ago. Unfortunately I couldn’t tell you exactly when, but the Pi stopped needing the file addition around 18 months ago. That’s around the time the kernel was bumped up to v 4 I think. Whether it is any sort of problem to have such a file when it isn’t required… I don’t know. It’s certainly less alarming for any user if the script doesn’t demand their superuser password!


tim
--
tim Rowledge; [hidden email]; http://www.rowledge.org/tim
Oxymorons: Good grief



Reply | Threaded
Open this post in threaded view
|

Re: [ANN] Squeak 5.1 Code Freeze -- Trunk still closed; first release candidate(s) available

fniephaus
On Fri, Aug 19, 2016 at 11:31 PM tim Rowledge <[hidden email]> wrote:

> On 19-08-2016, at 12:41 PM, Fabio Niephaus <[hidden email]> wrote:
>
> Thanks for the bug reports. I've fixed the regex bug and found a way to ensure "find" ignores dot files.

Excellent.

> But why do you think ensure_conf_file needs to depend upon the kernel level?

Because the issue was fixed in the kernel some time ago. Unfortunately I couldn’t tell you exactly when, but the Pi stopped needing the file addition around 18 months ago. That’s around the time the kernel was bumped up to v 4 I think. Whether it is any sort of problem to have such a file when it isn’t required… I don’t know. It’s certainly less alarming for any user if the script doesn’t demand their superuser password!

Can someone shed some light on this or do some more digging? When exactly do we need a squeak.conf? Any other ideas how to deal with this?

Fabio
 


tim
--
tim Rowledge; [hidden email]; http://www.rowledge.org/tim
Oxymorons: Good grief





Reply | Threaded
Open this post in threaded view
|

Re: [ANN] Squeak 5.1 Code Freeze -- Trunk still closed; first release candidate(s) available

Levente Uzonyi
On Fri, 19 Aug 2016, Fabio Niephaus wrote:

> On Fri, Aug 19, 2016 at 11:31 PM tim Rowledge <[hidden email]> wrote:
>
>       > On 19-08-2016, at 12:41 PM, Fabio Niephaus <[hidden email]> wrote:
>       >
>       > Thanks for the bug reports. I've fixed the regex bug and found a way to ensure "find" ignores dot files.
>
>       Excellent.
>
>       > But why do you think ensure_conf_file needs to depend upon the kernel level?
>
>       Because the issue was fixed in the kernel some time ago. Unfortunately I couldn’t tell you exactly when, but the Pi stopped needing the file addition around 18 months ago. That’s around the time
>       the kernel was bumped up to v 4 I think. Whether it is any sort of problem to have such a file when it isn’t required… I don’t know. It’s certainly less alarming for any user if the script doesn’t
>       demand their superuser password!
>
>
> Can someone shed some light on this or do some more digging? When exactly do we need a squeak.conf? Any other ideas how to deal with this?
You always need it when you use an ht VM. Without the file the VM will not
start (unless you run it as root, but that's something you wouldn't do,
would you?).
All VMs built by the CI are ht VMs, so the file is a must.
The file has no effect on kernels before 2.6.X (basically 10+ years old),
so checking the version is unnecessary (the VM would probably not start
anyway due to C library differences).

Levente

>
> Fabio
>  
>
>
>       tim
>       --
>       tim Rowledge; [hidden email]; http://www.rowledge.org/tim
>       Oxymorons: Good grief
>
>
>
>
>

Reply | Threaded
Open this post in threaded view
|

Re: [ANN] Squeak 5.1 Code Freeze -- Trunk still closed; first release candidate(s) available

timrowledge

> On 19-08-2016, at 4:32 PM, Levente Uzonyi <[hidden email]> wrote:
>
> On Fri, 19 Aug 2016, Fabio Niephaus wrote:
>
>> On Fri, Aug 19, 2016 at 11:31 PM tim Rowledge <[hidden email]> wrote:
>>
>>      > On 19-08-2016, at 12:41 PM, Fabio Niephaus <[hidden email]> wrote:
>>      >
>>      > Thanks for the bug reports. I've fixed the regex bug and found a way to ensure "find" ignores dot files.
>>
>>      Excellent.
>>
>>      > But why do you think ensure_conf_file needs to depend upon the kernel level?
>>
>>      Because the issue was fixed in the kernel some time ago. Unfortunately I couldn’t tell you exactly when, but the Pi stopped needing the file addition around 18 months ago. That’s around the time
>>      the kernel was bumped up to v 4 I think. Whether it is any sort of problem to have such a file when it isn’t required… I don’t know. It’s certainly less alarming for any user if the script doesn’t
>>      demand their superuser password!
>> Can someone shed some light on this or do some more digging? When exactly do we need a squeak.conf? Any other ideas how to deal with this?
>
> You always need it when you use an ht VM. Without the file the VM will not start (unless you run it as root, but that's something you wouldn't do, would you?).
> All VMs built by the CI are ht VMs, so the file is a must.


This is not strictly true; recent kernels simply don’t cause the problem. My Pi just doesn’t need the file and hasn’t for 18+ months. I’ve checked for it pre-existing and it doesn’t. I’ve been delivering the PI system with an ht based vm for a long time now and millions of people use it daily.


> The file has no effect on kernels before 2.6.X (basically 10+ years old), so checking the version is unnecessary (the VM would probably not start anyway due to C library differences).
>
> Levente
>
>> Fabio
>>  
>>
>>      tim
>>      --
>>      tim Rowledge; [hidden email]; http://www.rowledge.org/tim
>>      Oxymorons: Good grief
>


tim
--
tim Rowledge; [hidden email]; http://www.rowledge.org/tim
To define recursion, we must first define recursion.



Reply | Threaded
Open this post in threaded view
|

Re: [ANN] Squeak 5.1 Code Freeze -- Trunk still closed; first release candidate(s) available

marcel.taeumel
tim Rowledge wrote
> On 19-08-2016, at 4:32 PM, Levente Uzonyi <[hidden email]> wrote:
>
> On Fri, 19 Aug 2016, Fabio Niephaus wrote:
>
>> On Fri, Aug 19, 2016 at 11:31 PM tim Rowledge <[hidden email]> wrote:
>>
>>      > On 19-08-2016, at 12:41 PM, Fabio Niephaus <[hidden email]> wrote:
>>      >
>>      > Thanks for the bug reports. I've fixed the regex bug and found a way to ensure "find" ignores dot files.
>>
>>      Excellent.
>>
>>      > But why do you think ensure_conf_file needs to depend upon the kernel level?
>>
>>      Because the issue was fixed in the kernel some time ago. Unfortunately I couldn’t tell you exactly when, but the Pi stopped needing the file addition around 18 months ago. That’s around the time
>>      the kernel was bumped up to v 4 I think. Whether it is any sort of problem to have such a file when it isn’t required… I don’t know. It’s certainly less alarming for any user if the script doesn’t
>>      demand their superuser password!
>> Can someone shed some light on this or do some more digging? When exactly do we need a squeak.conf? Any other ideas how to deal with this?
>
> You always need it when you use an ht VM. Without the file the VM will not start (unless you run it as root, but that's something you wouldn't do, would you?).
> All VMs built by the CI are ht VMs, so the file is a must.


This is not strictly true; recent kernels simply don’t cause the problem. My Pi just doesn’t need the file and hasn’t for 18+ months. I’ve checked for it pre-existing and it doesn’t. I’ve been delivering the PI system with an ht based vm for a long time now and millions of people use it daily.


> The file has no effect on kernels before 2.6.X (basically 10+ years old), so checking the version is unnecessary (the VM would probably not start anyway due to C library differences).
>
> Levente
>
>> Fabio
>>  
>>
>>      tim
>>      --
>>      tim Rowledge; [hidden email]; http://www.rowledge.org/tim
>>      Oxymorons: Good grief
>


tim
--
tim Rowledge; [hidden email]; http://www.rowledge.org/tim
To define recursion, we must first define recursion.
Here is another release candidate:
http://files.squeak.org/5.1rc2/Squeak5.1rc2-16535-32bit/

The 64-bit VM for macOS we use on TravisCI has some hick-ups at the moment.

Best,
Marcel
Reply | Threaded
Open this post in threaded view
|

Re: [ANN] Squeak 5.1 Code Freeze -- Trunk still closed; first release candidate(s) available

fniephaus
In reply to this post by timrowledge
On Sat, Aug 20, 2016 at 1:46 AM tim Rowledge <[hidden email]> wrote:

> On 19-08-2016, at 4:32 PM, Levente Uzonyi <[hidden email]> wrote:
>
> On Fri, 19 Aug 2016, Fabio Niephaus wrote:
>
>> On Fri, Aug 19, 2016 at 11:31 PM tim Rowledge <[hidden email]> wrote:
>>
>>      > On 19-08-2016, at 12:41 PM, Fabio Niephaus <[hidden email]> wrote:
>>      >
>>      > Thanks for the bug reports. I've fixed the regex bug and found a way to ensure "find" ignores dot files.
>>
>>      Excellent.
>>
>>      > But why do you think ensure_conf_file needs to depend upon the kernel level?
>>
>>      Because the issue was fixed in the kernel some time ago. Unfortunately I couldn’t tell you exactly when, but the Pi stopped needing the file addition around 18 months ago. That’s around the time
>>      the kernel was bumped up to v 4 I think. Whether it is any sort of problem to have such a file when it isn’t required… I don’t know. It’s certainly less alarming for any user if the script doesn’t
>>      demand their superuser password!
>> Can someone shed some light on this or do some more digging? When exactly do we need a squeak.conf? Any other ideas how to deal with this?
>
> You always need it when you use an ht VM. Without the file the VM will not start (unless you run it as root, but that's something you wouldn't do, would you?).
> All VMs built by the CI are ht VMs, so the file is a must.


This is not strictly true; recent kernels simply don’t cause the problem. My Pi just doesn’t need the file and hasn’t for 18+ months. I’ve checked for it pre-existing and it doesn’t. I’ve been delivering the PI system with an ht based vm for a long time now and millions of people use it daily.

May I suggest we simply skip the ensure_conf_file step, if the launch script detects to run on a Pi for this release? Shall we print instructions for the common-session setup if necessary instead? We can adjust this behavior in the future, of course. But I'm not sure we can work out a proper solution in the next couple of days.

Fabio
 


> The file has no effect on kernels before 2.6.X (basically 10+ years old), so checking the version is unnecessary (the VM would probably not start anyway due to C library differences).
>
> Levente
>
>> Fabio
>>
>>
>>      tim
>>      --
>>      tim Rowledge; [hidden email]; http://www.rowledge.org/tim
>>      Oxymorons: Good grief
>


tim
--
tim Rowledge; [hidden email]; http://www.rowledge.org/tim
To define recursion, we must first define recursion.





Reply | Threaded
Open this post in threaded view
|

Re: [ANN] Squeak 5.1 Code Freeze -- Trunk still closed; first release candidate(s) available

Hannes Hirzel
On 8/20/16, Fabio Niephaus <[hidden email]> wrote:

> On Sat, Aug 20, 2016 at 1:46 AM tim Rowledge <[hidden email]> wrote:
>
>>
>> > On 19-08-2016, at 4:32 PM, Levente Uzonyi <[hidden email]> wrote:
>> >
>> > On Fri, 19 Aug 2016, Fabio Niephaus wrote:
>> >
>> >> On Fri, Aug 19, 2016 at 11:31 PM tim Rowledge <[hidden email]>
>> >> wrote:
>> >>
>> >>      > On 19-08-2016, at 12:41 PM, Fabio Niephaus
>> >> <[hidden email]>
>> wrote:
>> >>      >
>> >>      > Thanks for the bug reports. I've fixed the regex bug and found
>> >> a
>> way to ensure "find" ignores dot files.
>> >>
>> >>      Excellent.
>> >>
>> >>      > But why do you think ensure_conf_file needs to depend upon the
>> kernel level?
>> >>
>> >>      Because the issue was fixed in the kernel some time ago.
>> Unfortunately I couldn’t tell you exactly when, but the Pi stopped
>> needing
>> the file addition around 18 months ago. That’s around the time
>> >>      the kernel was bumped up to v 4 I think. Whether it is any sort
>> >> of
>> problem to have such a file when it isn’t required… I don’t know. It’s
>> certainly less alarming for any user if the script doesn’t
>> >>      demand their superuser password!
>> >> Can someone shed some light on this or do some more digging? When
>> exactly do we need a squeak.conf? Any other ideas how to deal with this?
>> >
>> > You always need it when you use an ht VM. Without the file the VM will
>> not start (unless you run it as root, but that's something you wouldn't
>> do,
>> would you?).
>> > All VMs built by the CI are ht VMs, so the file is a must.
>>
>>
>> This is not strictly true; recent kernels simply don’t cause the problem.
>> My Pi just doesn’t need the file and hasn’t for 18+ months. I’ve checked
>> for it pre-existing and it doesn’t. I’ve been delivering the PI system
>> with
>> an ht based vm for a long time now and millions of people use it daily.
>>
>
> May I suggest we simply skip the ensure_conf_file step, if the launch
> script detects to run on a Pi for this release?
+1

Shall we print instructions

> for the common-session setup if necessary instead? We can adjust this
> behavior in the future, of course. But I'm not sure we can work out a
> proper solution in the next couple of days.
>
> Fabio
>
>
>>
>>
>> > The file has no effect on kernels before 2.6.X (basically 10+ years
>> old), so checking the version is unnecessary (the VM would probably not
>> start anyway due to C library differences).
>> >
>> > Levente
>> >
>> >> Fabio
>> >>
>> >>
>> >>      tim
>> >>      --
>> >>      tim Rowledge; [hidden email]; http://www.rowledge.org/tim
>> >>      Oxymorons: Good grief
>> >
>>
>>
>> tim
>> --
>> tim Rowledge; [hidden email]; http://www.rowledge.org/tim
>> To define recursion, we must first define recursion.
>>
>>
>>
>>
>

Reply | Threaded
Open this post in threaded view
|

Re: [ANN] Squeak 5.1 Code Freeze -- Trunk still closed; first release candidate(s) available

Chris Muller-3
In reply to this post by fniephaus
> I’ve been delivering the PI system with an
>> ht based vm for a long time now and millions of people use it daily.

"Millions?"  You've really piqued my interest Fabio!  Can you share
any info?  :)

Reply | Threaded
Open this post in threaded view
|

Re: [ANN] Squeak 5.1 Code Freeze -- Trunk still closed; first release candidate(s) available

timrowledge

> On 20-08-2016, at 9:57 AM, Chris Muller <[hidden email]> wrote:
>
>> I’ve been delivering the PI system with an
>>> ht based vm for a long time now and millions of people use it daily.
>
> "Millions?"  You've really piqued my interest Fabio!  Can you share
> any info?  :)
>

Err, that would be me, referring to Scratch on Pi.

tim
--
tim Rowledge; [hidden email]; http://www.rowledge.org/tim
Strange OpCodes: XYZZY: Branch and Play Adventure



Reply | Threaded
Open this post in threaded view
|

Re: [ANN] Squeak 5.1 Code Freeze -- Trunk still closed; first release candidate(s) available

Chris Muller-4
Oh.  Just you, nevermind..    ;-)

On Sat, Aug 20, 2016 at 12:08 PM, tim Rowledge <[hidden email]> wrote:

>
>> On 20-08-2016, at 9:57 AM, Chris Muller <[hidden email]> wrote:
>>
>>> I’ve been delivering the PI system with an
>>>> ht based vm for a long time now and millions of people use it daily.
>>
>> "Millions?"  You've really piqued my interest Fabio!  Can you share
>> any info?  :)
>>
>
> Err, that would be me, referring to Scratch on Pi.
>
> tim
> --
> tim Rowledge; [hidden email]; http://www.rowledge.org/tim
> Strange OpCodes: XYZZY: Branch and Play Adventure
>
>

Reply | Threaded
Open this post in threaded view
|

Re: [ANN] Squeak 5.1 Code Freeze -- Trunk still closed; first release candidate(s) available

timrowledge
In reply to this post by fniephaus

> On 20-08-2016, at 3:46 AM, Fabio Niephaus <[hidden email]> wrote:
>

> May I suggest we simply skip the ensure_conf_file step, if the launch script detects to run on a Pi for this release? Shall we print instructions for the common-session setup if necessary instead? We can adjust this behavior in the future, of course. But I'm not sure we can work out a proper solution in the next couple of days.

I would suggest that you make the ensure_conf_file script start with a test for

a) being on a Pi with a kernel level 4 or greater
b) being on any machine with kernel greater than X, where X is not currently a number I know. I simply don’t know enough linux terminology to do effective googling for this. The Pi does not have a special kernel so I think any version 4+ would be safe. Is anyone running on an x86 linux out there?

If NOT (a AND b) then do magic with /etc/securityblah-blah.

Somebody that follows and understand linux release history would probably be able to tell us exactly which version introduced the fix that makes /etc/security/blah unnecessary.

A test for /etc/pam.d/common-session containing
'       session    required   pam_limits.so'
would be a decent idea if we could also work out if the user were using xrdp or any other affected system. I’d suggest just documenting it for now.

tim
--
tim Rowledge; [hidden email]; http://www.rowledge.org/tim
Strange OpCodes: BFM: Branch on Full Moon



Reply | Threaded
Open this post in threaded view
|

Re: [ANN] Squeak 5.1 Code Freeze -- Trunk still closed; first release candidate(s) available

timrowledge

> On 20-08-2016, at 10:38 AM, tim Rowledge <[hidden email]> wrote:
>

> If NOT (a AND b) then do magic with /etc/securityblah-blah.

Sigh. if NOT( a OR  b)

tim
--
tim Rowledge; [hidden email]; http://www.rowledge.org/tim
Java:  the best argument for Smalltalk since C++



Reply | Threaded
Open this post in threaded view
|

Re: [ANN] Squeak 5.1 Code Freeze -- Trunk still closed; first release candidate(s) available

fniephaus
In reply to this post by timrowledge
On Sat, Aug 20, 2016 at 7:38 PM tim Rowledge <[hidden email]> wrote:

> On 20-08-2016, at 3:46 AM, Fabio Niephaus <[hidden email]> wrote:
>

> May I suggest we simply skip the ensure_conf_file step, if the launch script detects to run on a Pi for this release? Shall we print instructions for the common-session setup if necessary instead? We can adjust this behavior in the future, of course. But I'm not sure we can work out a proper solution in the next couple of days.

I would suggest that you make the ensure_conf_file script start with a test for

a) being on a Pi with a kernel level 4 or greater
b) being on any machine with kernel greater than X, where X is not currently a number I know. I simply don’t know enough linux terminology to do effective googling for this. The Pi does not have a special kernel so I think any version 4+ would be safe. Is anyone running on an x86 linux out there?

If NOT (a AND b) then do magic with /etc/securityblah-blah.

Somebody that follows and understand linux release history would probably be able to tell us exactly which version introduced the fix that makes /etc/security/blah unnecessary.

I just had a quick chat with Eliot and we decided to skip the ensure_conf_file step on systems with Linux kernels earlier than 4.x.x for now.
We can change the behavior in the next release, but I'd currently say, it works well on the different systems as is.

Best,
Fabio


A test for /etc/pam.d/common-session containing
'       session    required   pam_limits.so'
would be a decent idea if we could also work out if the user were using xrdp or any other affected system. I’d suggest just documenting it for now.

tim
--
tim Rowledge; [hidden email]; http://www.rowledge.org/tim
Strange OpCodes: BFM: Branch on Full Moon