Re: On backporting fix for tdf#124503 "LibreOffice doesn't detect JVM because of unexpected java.vendor property value"

2019-04-13 Thread rene . engelhard 
Am 12. April 2019 09:43:01 MESZ schrieb Stephan Bergmann :
>On 11/04/2019 10:35, Stephan Bergmann wrote:
>> On 11/04/2019 09:30, Thorsten Behrens wrote:
>>> I'd prefer a backport to both maintained branches - at the current
>>> speed of new binary Java distros springing up (and the overall
>>> trajectory of that ecosystem there), I'd expect more of those issues
>>> down the road.
>> 
>> Fair enough.  Gerrit changes up for review at 
>>  for libreoffice-6-1 and
>at 
>>  for libreoffice-6-2.
>
>(BTW, I only tested my fix against stock OpenJDK reporting the Oracle 
>vendor string, by manually removing that vendor information from my 
>local LO sources temporarily.  It would be great if somebody could 
>actually test it against one of those problematic OpenJDK variants that
>
>are unknown in the LO sources.)
>___
>LibreOffice mailing list
>LibreOffice@lists.freedesktop.org
>https://lists.freedesktop.org/mailman/listinfo/libreoffice

Hi,

Did that last night with the libreoffice-6-1 branch including the backport.

Seems to work. At least no test failure :)

Regards,

Rene
-- 
Diese Nachricht wurde von meinem Android-Gerät mit K-9 Mail gesendet.
___
LibreOffice mailing list
LibreOffice@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/libreoffice

Re: On backporting fix for tdf#124503 "LibreOffice doesn't detect JVM because of unexpected java.vendor property value"

2019-04-12 Thread Stephan Bergmann 

On 11/04/2019 10:35, Stephan Bergmann wrote:

On 11/04/2019 09:30, Thorsten Behrens wrote:

I'd prefer a backport to both maintained branches - at the current
speed of new binary Java distros springing up (and the overall
trajectory of that ecosystem there), I'd expect more of those issues
down the road.


Fair enough.  Gerrit changes up for review at 
 for libreoffice-6-1 and at 
 for libreoffice-6-2.


(BTW, I only tested my fix against stock OpenJDK reporting the Oracle 
vendor string, by manually removing that vendor information from my 
local LO sources temporarily.  It would be great if somebody could 
actually test it against one of those problematic OpenJDK variants that 
are unknown in the LO sources.)

___
LibreOffice mailing list
LibreOffice@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/libreoffice

Re: On backporting fix for tdf#124503 "LibreOffice doesn't detect JVM because of unexpected java.vendor property value"

2019-04-11 Thread Stephan Bergmann 

On 11/04/2019 09:30, Thorsten Behrens wrote:

I'd prefer a backport to both maintained branches - at the current
speed of new binary Java distros springing up (and the overall
trajectory of that ecosystem there), I'd expect more of those issues
down the road.


Fair enough.  Gerrit changes up for review at 
 for libreoffice-6-1 and at 
 for libreoffice-6-2.

___
LibreOffice mailing list
LibreOffice@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/libreoffice

Re: On backporting fix for tdf#124503 "LibreOffice doesn't detect JVM because of unexpected java.vendor property value"

2019-04-11 Thread Thorsten Behrens 
Hi Stephan,

Stephan Bergmann wrote:
> If not, the question is whether to backport the above fix to
> libreoffice-6-1 (towards LO 6.1.6), libreoffice-6-2 (towards LO
> 6.2.4), and maybe even libreoffice-6-2-3. The fix isn't exactly
> small, so I would prefer to not backport it aggressively.  But I
> don't know how severely users would be affected by this issue.
>
I'd prefer a backport to both maintained branches - at the current
speed of new binary Java distros springing up (and the overall
trajectory of that ecosystem there), I'd expect more of those issues
down the road.

Cheers,

-- Thorsten


signature.asc
Description: PGP signature
___
LibreOffice mailing list
LibreOffice@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/libreoffice

Re: On backporting fix for tdf#124503 "LibreOffice doesn't detect JVM because of unexpected java.vendor property value"

2019-04-10 Thread Olivier Tilloy 
On Wed, Apr 10, 2019 at 10:22 AM Stephan Bergmann 
wrote:

> See  for
> the fix for master (towards LO 6.3).  While the title and fix both cover
> a broad scope (any JRE that reports a java.vendor that the LO code
> doesn't find in its hardcoded list), the issue that prompted the bug and
> the fix is that Debian and Ubuntu apparently started to distribute
> OpenJDK versions that no longer announce the well-known Oracle
> java.vendor string, but instead go with things like "Debian", "Ubuntu",
> or "Private".
>
> It is not clear to me whether those distros will revert their
> modifications soonish (so that there would be no immediate need for LO
> to get anything fixed on our side).  If not, the question is whether to
> backport the above fix to libreoffice-6-1 (towards LO 6.1.6),
> libreoffice-6-2 (towards LO 6.2.4), and maybe even libreoffice-6-2-3.
> The fix isn't exactly small, so I would prefer to not backport it
> aggressively.  But I don't know how severely users would be affected by
> this issue.  (I assume that Debian and Ubuntu would take care of the
> issue for their bundled LO, by updating the hard-coded list accordingly.
>   That could also be an alternative to backporting the above fix here at
> upstream, but with drawbacks:  We would---somewhat needlessly---extend
> the hardcoded list, even if master already has a fix that makes the
> additions moot.)
>
> Thoughts, esp. from people involved in the relevant distros?


There doesn't seem to be an intention to revert the java.vendor
modifications in Debian and Ubuntu.
Your assumption that this is distro-patched in Debian and Ubuntu is
correct, see
https://salsa.debian.org/libreoffice-team/libreoffice/libreoffice/blob/debian-experimental-6.2/patches/java.vendor-Debian.diff
 and
https://git.launchpad.net/~libreoffice/ubuntu/+source/libreoffice/tree/patches/java.vendor-Ubuntu.patch?h=ubuntu-disco-6.2
.
So there's not immediate need for backporting the fix from master to 6.2.x.
If you do that would allow dropping those two distro patches.

Thanks for addressing the issue in master so promptly, by the way!

Cheers,

 Olivier
___
LibreOffice mailing list
LibreOffice@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/libreoffice

On backporting fix for tdf#124503 "LibreOffice doesn't detect JVM because of unexpected java.vendor property value"

2019-04-10 Thread Stephan Bergmann 
See  for 
the fix for master (towards LO 6.3).  While the title and fix both cover 
a broad scope (any JRE that reports a java.vendor that the LO code 
doesn't find in its hardcoded list), the issue that prompted the bug and 
the fix is that Debian and Ubuntu apparently started to distribute 
OpenJDK versions that no longer announce the well-known Oracle 
java.vendor string, but instead go with things like "Debian", "Ubuntu", 
or "Private".


It is not clear to me whether those distros will revert their 
modifications soonish (so that there would be no immediate need for LO 
to get anything fixed on our side).  If not, the question is whether to 
backport the above fix to libreoffice-6-1 (towards LO 6.1.6), 
libreoffice-6-2 (towards LO 6.2.4), and maybe even libreoffice-6-2-3. 
The fix isn't exactly small, so I would prefer to not backport it 
aggressively.  But I don't know how severely users would be affected by 
this issue.  (I assume that Debian and Ubuntu would take care of the 
issue for their bundled LO, by updating the hard-coded list accordingly. 
 That could also be an alternative to backporting the above fix here at 
upstream, but with drawbacks:  We would---somewhat needlessly---extend 
the hardcoded list, even if master already has a fix that makes the 
additions moot.)


Thoughts, esp. from people involved in the relevant distros?
___
LibreOffice mailing list
LibreOffice@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/libreoffice