"AtkWrapper not found" error impacting JOSM on Ubuntu/Mint

classic Classic list List threaded Threaded
6 messages Options
Reply | Threaded
Open this post in threaded view
|

"AtkWrapper not found" error impacting JOSM on Ubuntu/Mint

Vincent Privat-2
Hello,
Since today we are spammed by bug reports from Linux Mint and Ubuntu when
people try to launch JOSM with the latest Java update
(10.0.2+13-Ubuntu-1ubuntu0.18.04.1):

https://josm.openstreetmap.de/ticket/16666

The error is "ClassNotFoundException: org.GNOME.Accessibility.AtkWrapper".

Is it a known issue? We cannot really workaround it unless telling every
impacted user to manually disable ATK wrapper in accessibility.properties.

Thanks,
Vincent
Reply | Threaded
Open this post in threaded view
|

Re: "AtkWrapper not found" error impacting JOSM on Ubuntu/Mint

Vincent Privat-2
I found this bug:
https://bugs.launchpad.net/ubuntu/+source/openjdk-lts/+bug/1788250
Vincent

Le mar. 21 août 2018 à 23:27, Vincent Privat <[hidden email]>
a écrit :

> Hello,
> Since today we are spammed by bug reports from Linux Mint and Ubuntu when
> people try to launch JOSM with the latest Java update
> (10.0.2+13-Ubuntu-1ubuntu0.18.04.1):
>
> https://josm.openstreetmap.de/ticket/16666
>
> The error is "ClassNotFoundException: org.GNOME.Accessibility.AtkWrapper".
>
> Is it a known issue? We cannot really workaround it unless telling every
> impacted user to manually disable ATK wrapper in accessibility.properties.
>
> Thanks,
> Vincent
>
Reply | Threaded
Open this post in threaded view
|

Re: "AtkWrapper not found" error impacting JOSM on Ubuntu/Mint

Vincent Privat-2
I remember too, it was indeed a nightmare and I don't want to live that
again.
It took me a while but I found a workaround that doesn't need any user
action nor any adherence to ATK wrapper.
We can just catch the AWTError and try again to re-create our MainFrame. As
the ATK initialization fails only after the AWT toolkit has been created,
the second call succeeds:
https://josm.openstreetmap.de/changeset/14177/josm

Kevin, can you please check tomorrow that it works for you too?
I'm leaving the ticket open a few days as we will probably receive many
more bug reports this week.

Right now it's too difficult to disable ATK on Java 10 and remain
compatible with Java 8. I'll revisit this topic once we migrate to Java 11.

Cheers,
Vincent

Le mer. 22 août 2018 à 01:14, Kevin Kenny <[hidden email]> a
écrit :

> On Tue, Aug 21, 2018 at 5:48 PM Vincent Privat
> <[hidden email]> wrote:
> >
> > I found this bug:
> > https://bugs.launchpad.net/ubuntu/+source/openjdk-lts/+bug/1788250
> > Vincent
> >
> > Le mar. 21 août 2018 à 23:27, Vincent Privat <
> [hidden email]>
> > a écrit :
> >
> > > Hello,
> > > Since today we are spammed by bug reports from Linux Mint and Ubuntu
> when
> > > people try to launch JOSM with the latest Java update
> > > (10.0.2+13-Ubuntu-1ubuntu0.18.04.1):
> > >
> > > https://josm.openstreetmap.de/ticket/16666
> > >
> > > The error is "ClassNotFoundException:
> org.GNOME.Accessibility.AtkWrapper".
> > >
> > > Is it a known issue? We cannot really workaround it unless telling
> every
> > > impacted user to manually disable ATK wrapper in
> accessibility.properties.
> > >
> > > Thanks,
> > > Vincent
> > >
>
> Even deleting the ATK wrapper in accessibility.properties doesn't help,
> you also have to launch JOSM with a command like
>
> java -Xmx1g -Xbootclasspath/a:/usr/share/java/java-atk-wrapper.jar
> -jar ~/josm/josm-tested.jar
>
> When I do both things, it appears to work around the problem.
>
> AtkWrapper is a nightmare. I remember when
> https://josm.openstreetmap.de/ticket/12022
> had me using Meerkartor for a while.
>
Reply | Threaded
Open this post in threaded view
|

Re: "AtkWrapper not found" error impacting JOSM on Ubuntu/Mint

Vincent Privat-2
JOSM 14178 has been released as new tested version to workaround this
problem.
Samuel, do you know when the root issue will be fixed in Java ATK wrapper?
Has it ever been considered that this package is made optional rather than
a strict dependency of openjdk-<n>-jre? Most users don't need it.
Cheers,
Vincent

Le mer. 22 août 2018 à 02:49, Vincent Privat <[hidden email]>
a écrit :

> I remember too, it was indeed a nightmare and I don't want to live that
> again.
> It took me a while but I found a workaround that doesn't need any user
> action nor any adherence to ATK wrapper.
> We can just catch the AWTError and try again to re-create our MainFrame.
> As the ATK initialization fails only after the AWT toolkit has been
> created, the second call succeeds:
> https://josm.openstreetmap.de/changeset/14177/josm
>
> Kevin, can you please check tomorrow that it works for you too?
> I'm leaving the ticket open a few days as we will probably receive many
> more bug reports this week.
>
> Right now it's too difficult to disable ATK on Java 10 and remain
> compatible with Java 8. I'll revisit this topic once we migrate to Java 11.
>
> Cheers,
> Vincent
>
> Le mer. 22 août 2018 à 01:14, Kevin Kenny <[hidden email]> a
> écrit :
>
>> On Tue, Aug 21, 2018 at 5:48 PM Vincent Privat
>> <[hidden email]> wrote:
>> >
>> > I found this bug:
>> > https://bugs.launchpad.net/ubuntu/+source/openjdk-lts/+bug/1788250
>> > Vincent
>> >
>> > Le mar. 21 août 2018 à 23:27, Vincent Privat <
>> [hidden email]>
>> > a écrit :
>> >
>> > > Hello,
>> > > Since today we are spammed by bug reports from Linux Mint and Ubuntu
>> when
>> > > people try to launch JOSM with the latest Java update
>> > > (10.0.2+13-Ubuntu-1ubuntu0.18.04.1):
>> > >
>> > > https://josm.openstreetmap.de/ticket/16666
>> > >
>> > > The error is "ClassNotFoundException:
>> org.GNOME.Accessibility.AtkWrapper".
>> > >
>> > > Is it a known issue? We cannot really workaround it unless telling
>> every
>> > > impacted user to manually disable ATK wrapper in
>> accessibility.properties.
>> > >
>> > > Thanks,
>> > > Vincent
>> > >
>>
>> Even deleting the ATK wrapper in accessibility.properties doesn't help,
>> you also have to launch JOSM with a command like
>>
>> java -Xmx1g -Xbootclasspath/a:/usr/share/java/java-atk-wrapper.jar
>> -jar ~/josm/josm-tested.jar
>>
>> When I do both things, it appears to work around the problem.
>>
>> AtkWrapper is a nightmare. I remember when
>> https://josm.openstreetmap.de/ticket/12022
>> had me using Meerkartor for a while.
>>
>
Reply | Threaded
Open this post in threaded view
|

Re: "AtkWrapper not found" error impacting JOSM on Ubuntu/Mint

Vincent Privat-2
Hello,
The bug is closed so there is nothing to expect from OpenJDK. The solution
must come from Debian or a new bug created.
The Java version which introduced the regression is
10.0.2+13-Ubuntu-1ubuntu0.18.04.1
because it contains the fix to move accessibility.properties in the "conf"
folder. The issue did not occur before because the configuration file
enabling ATK wrapper was not found by openjdk. A few hours after this
version was delivered by Ubuntu we started to receive a dozen of bug
reports, until I delivered a JOSM hotfix mitigating the issue.

Patching openjdk with your try/catch proposal and making the ATK wrapper a
Recommends sounds a good idea.

Don't wait for openjdk guys for an answer: they simply don't care anymore
with desktop technologies.

Cheers,
Vincent

Le ven. 24 août 2018 à 11:47, Samuel Thibault <[hidden email]> a
écrit :

> Hello,
>
> Vincent Privat, le mer. 22 août 2018 21:01:16 +0200, a ecrit:
> > Samuel, do you know when the root issue will be fixed in Java ATK
> wrapper?
>
> The root issue is in openjdk, not the wrapper. Openjdk has apparently
> removed the way that was used to load the wrapper, without apparently
> making sure that another way was implemented. I have forwarded the issue
> upstream
>
> http://bugs.java.com/bugdatabase/view_bug.do?bug_id=JDK-8204862
>
> but apparently they don't have the issue (I guess they don't actually
> manage to enable the wrapper in the configuration file).
>
> > > > Since today we are spammed by bug reports from Linux Mint and Ubuntu
> when
> > > > people try to launch JOSM with the latest Java update
> > > > (10.0.2+13-Ubuntu-1ubuntu0.18.04.1):
>
> Which version is introducing the issue exactly?  AFAIK it's the whole
> openjdk >= 9 which poses the problem.
>
> > Has it ever been considered that this package is made optional rather
> than a
> > strict dependency of openjdk-<n>-jre?
>
> The problem is not the dependency, but what openjdk itself is doing. ATM
> it seems to be insisting on loading the wrapper even when it's not
> available.  I guess catching exceptions around
> src/java.desktop/share/classes/java/awt/Toolkit.java's getDefaultToolkit
> call to loadAssistiveTechnologies, to just print the exception but not
> abort, would allow to make the dependency a Recommends only.
>
> Concerning actually managing to load the wrapper, see the backlog of
> #900912. I have commited to the git repo of java-atk-wrapper what is
> said to be needed for making it loadable with openjdk 9 and later, but
> as I said there, we still seem to be lacking in openjdk a path where the
> wrapper can be put for openjdk to find it. And it's been a month without
> anybody answering.
>
> Samuel
>
Reply | Threaded
Open this post in threaded view
|

Re: Enabling jaw (Java-atk-wrapper) by default ? (Bug#900912)

Vincent Privat-2
If enabled by default, please offer a reliable way for applications to
disable it. We don't need it for JOSM, and we have been so impacted with
jaw's problems in the past years that we will never want it enabled by
default for us.

Le dim. 7 avr. 2019 à 12:08, Samuel Thibault <[hidden email]> a
écrit :

> Hello,
>
> Matthias Klose, le sam. 06 avril 2019 15:46:21 +0200, a ecrit:
> > On 06.04.19 15:13, Paul Gevers wrote:
> > > We're late already, I would want this rather sooner than latter
> > > in buster, such that there is some real live testing before we release.
> > > Sure, there are chances for bugs, but if that's the case, let's find
> > > them and fix them.
> >
> > I disagree.  I'll do the next upload with Samuel's proposed patches, not
> > enabling that by default, together with the planned security update.
> Then
> > people can start testing if the wrapper works.
>
> Well, I'm afraid that what will happen is that the people who will
> test will simply find out that it just works for them (just like it
> does already for them with openjdk-8) ; will we then conclude near the
> release that it should be enabled by default for Buster, and then be hit
> by the much wider exposition to jaw?
>
> If on the contrary we enable it by default during the freeze, we will
> have *way* more testing coverage, and thus be much more confident with
> keeping it enabled by default in Buster if we don't see bug reports.
>
> > Enabling features during the freeze which were broken most of the time
> > during the development cycle sounds risky.
>
> Just ftr: what was broken was the load of jaw in openjdk-11, jaw itself
> seems to work in openjdk-8 for people needing it.
>
> Samuel
>