Pre-release of Geppetto 4.2 available for testing

[Thomas Hallgren]

The work on Geppetto 4.2 is feature complete and the artifacts from the latest build is available for anyone who would like to give it a test drive. Feedback is of course very much appreciated. The most notably changes in this release are:

• New text based editor for metadata.json with code-completion, documentation tooltips, quick-fixes and more.
• Target platform upgrade. Geppetto 4.2 is based on Eclipse Luna (4.3).
• Wizard structure changes. Import from Forge is now found under Import.
• Support for new all new metadata attributes
• Checksums stored in separate checksums.json file

Use this URL to get a full list of bugs fixed.

https://tickets.puppetlabs.com?jql=project%20%3D%20GEP%20AND%20fixVersion%20%3D%20%224.2.0

The pre-release artifacts can be found here:

http://geppetto-jenkins.puppetlabs.com/job/Geppetto%20Dev/lastSuccessfulBuild/artifact/

Feedback can be given either in this forum or as JIRA tickets here:

https://tickets.puppetlabs.com/browse/GEP

Please note that Geppetto 4.2 is based on Eclipse Luna (4.3). Those of you who would like to install it into an Eclipse IDE should be aware that installing it into older versions might well work but it has not been tested.

Regards,
Thomas Hallgren

[Pete Brown]

Awesome!

What is the best way to install this in Eclipse?
I have just installed Eclipse 4.4 here and the Gepetto 4.1 seems to be
having some issues.

> --
> You received this message because you are subscribed to the Google Groups
> "Puppet Geppetto" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to puppet-geppet...@googlegroups.com.
> For more options, visit https://groups.google.com/d/optout.

[Thomas Hallgren]

You'll find two links when using the link to the last successful artifacts:

http://geppetto-jenkins.puppetlabs.com/job/Geppetto%20Dev/lastSuccessfulBuild/artifact/

The one named 'geppetto.p2.repository', i.e.

http://geppetto-jenkins.puppetlabs.com/job/Geppetto%20Dev/lastSuccessfulBuild/artifact/geppetto.p2.repository

can be entered as a "Software Site" in the Eclipse installer. Adding that site and then doing a 'Check for updates'
should do the trick. If it doesn't work, then try first uninstalling Geppetto 4.1.

- thomas

[Thomas Hallgren]

The link for viewing the JIRA tickets was wrong. Please use this one instead:

  https://tickets.puppetlabs.com/issues/?jql=project%20%3D%20GEP%20AND%20fixVersion%20%3D%20%224.2.0%22%20ORDER%20BY%20key%20ASC

- thomas

[Pete Brown]

I just tried that and got this error.

Unable to read repository at
file:/Applications/Eclipse/p2/org.eclipse.equinox.p2.repository/cache/compositeContent-1408555024.xml

The element type "hr" must be terminated by the matching end-tag "</hr>".

[Thomas Hallgren]

Interesting. The Eclipse installer apparently encodes the URL so %20 in a URL doesn't work. As a workaround, try using:

http://geppetto-jenkins.puppetlabs.com/job/Geppetto Dev/lastSuccessfulBuild/artifact/geppetto.p2.repository

i.e. use a normal space instead of %20. That worked for me. Meanwhile I'll rename the job to something that doesn't have
spaces in it and report the Bug to Eclipse.

- thomas

[Thomas Hallgren]

I created a new job without spaces in the name so now you can use this URL:

http://geppetto-jenkins.puppetlabs.com/job/GeppettoDev/lastSuccessfulBuild/artifact/geppetto.p2.repository/

- thomas


On 2014-07-16 11:11, Pete Brown wrote:

[Pete Brown]

Excellent.
Thanks Thomas!

I will run it in as my normal dev environment and see how it goes.

[Pete Brown]

Has Geppetto's search tree been changed in 4.2?
I just loaded up 4.2 into my Eclipse 4.4 install and it seems to be
stuck in a loop building puppet projects.
It seems to be cycling through a couple of modules but I also have a
few r10k trees i use for testing and I am wondering if it has found
those and is searching through them.

[Thomas Hallgren]

On 2014-07-17 02:10, Pete Brown wrote:
> Has Geppetto's search tree been changed in 4.2?

In some sense, yes. The metadata.json is now built by a different type of builder which uses a different search algorithm.

> I just loaded up 4.2 into my Eclipse 4.4 install and it seems to be
> stuck in a loop building puppet projects.
> It seems to be cycling through a couple of modules but I also have a
> few r10k trees i use for testing and I am wondering if it has found
> those and is searching through them.

Can you try closing one of those projects and see if that makes a difference? In any case, it would be very interesting
to, if possible, try your workspace (or something that exhibits the same problem).

- thomas

[Pete Brown]

On 17 July 2014 17:47, Thomas Hallgren <thomas....@puppetlabs.com> wrote:
> On 2014-07-17 02:10, Pete Brown wrote:
>>
>> Has Geppetto's search tree been changed in 4.2?
>
> In some sense, yes. The metadata.json is now built by a different type of
> builder which uses a different search algorithm.

This is the major feature I am awaiting.
Thanks for adding it.

>> I just loaded up 4.2 into my Eclipse 4.4 install and it seems to be
>> stuck in a loop building puppet projects.
>> It seems to be cycling through a couple of modules but I also have a
>> few r10k trees i use for testing and I am wondering if it has found
>> those and is searching through them.
>
> Can you try closing one of those projects and see if that makes a
> difference? In any case, it would be very interesting to, if possible, try
> your workspace (or something that exhibits the same problem).

It turns out it was looking through all the old build directories in
the pkg directory for my puppet modules.
It wasn't behaving the same way with a separate workspace I use for a
clients internal modules that never get published to puppet forge.

Thanks for the help.

>
> - thomas
>
>
>> On 17 July 2014 08:08, Pete Brown <rendh...@gmail.com> wrote:
>>>
>>> Excellent.
>>> Thanks Thomas!
>>>
>>> I will run it in as my normal dev environment and see how it goes.
>>>
>>> On 17 July 2014 01:28, Thomas Hallgren <thomas....@puppetlabs.com>
>>> wrote:
>>>
>

[Pete Brown]

I have been thinking about this.
Geppetto was definitely searching through the pkg and spec trees for
puppet manifests.
I had a bunch of builds in my pkg directories and my spec directories
contained symlinks to the root directory of the module.
The symlinks were obviously causing some seriously recursive searching
for manifests.
This was obviously causing issues for me and I am wondering if the pkg
and spec directories should be ignored for updates to the workspace.

[Thomas Hallgren]

On 2014-07-18 08:02, Pete Brown wrote:
> I have been thinking about this.
> Geppetto was definitely searching through the pkg and spec trees for
> puppet manifests.
> I had a bunch of builds in my pkg directories and my spec directories
> contained symlinks to the root directory of the module.
> The symlinks were obviously causing some seriously recursive searching
> for manifests.
> This was obviously causing issues for me and I am wondering if the pkg
> and spec directories should be ignored for updates to the workspace.

That sounds reasonable. And perhaps the set of ignored directories
should be configurable?

Can you please enter a JIRA issue for this:
https://tickets.puppetlabs.com/browse/GEP

Thanks,

[Simon Tremblay]

Thank you very much for that update. 

It is working very well here with Eclipse Luna 4.4.0.

[Brian Keyser]

Hi Thomas,

I have been having difficulties using the p2 repository.  I have adjusted the download site to not include the space, but the zip files that are downloads are corrupted.  For example, this is the content.jar file in the p2 cache

$ jar xf content475293631.jar

java.util.zip.ZipException: invalid distance too far back

at java.util.zip.InflaterInputStream.read(InflaterInputStream.java:164)

at java.util.zip.ZipInputStream.read(ZipInputStream.java:193)

at java.util.zip.ZipInputStream.closeEntry(ZipInputStream.java:139)

at sun.tools.jar.Main.extractFile(Main.java:961)

at sun.tools.jar.Main.extract(Main.java:877)

at sun.tools.jar.Main.run(Main.java:263)

at sun.tools.jar.Main.main(Main.java:1177)

Jenkins suggests that there is a problem with downloading zip files in JENKINS-19752.  However, I have noticed that several others have been able to use the p2 site without issue (once the space was addressed).  Perhaps Jenkins was recently updated to a version that includes this defect?

Thanks,

Brian 

[Thomas Hallgren]

Hi Brian,

I'm using both the new URL (space removed):

http://geppetto-jenkins.puppetlabs.com/job/GeppettoDev/lastSuccessfulBuild/artifact/geppetto.p2.repository/

and the old one:

http://geppetto-jenkins.puppetlabs.com/job/Geppetto Dev/lastSuccessfulBuild/artifact/geppetto.p2.repository/

without problems.  Neither have changed since July 16. Could it be that you had some network glitch when downloading and therefore have a corrupt file in your cache? Please try reloading the update site and see if that fixes the problem (click on "Available Software sites" in the installer, select the site URL and then click "Reload").

HTH,
- thomas

[David Schmitt]

Hi,

I've got the same error:

> Unable to read repository at

> file:/C:/Users/schmidav/Downloads/eclipse/p2/org.eclipse.equinox.p2.repository/cache/content475293631.jar.

> Unable to read repository at

> file:/C:/Users/schmidav/Downloads/eclipse/p2/org.eclipse.equinox.p2.repository/cache/content475293631.jar.

> invalid distance too far back

This is a fresh eclipse 4.4 install on windows 7.

MD5SUM:
7e3939c0c182c90f1a4044181b5454f3
*/c/Users/schmidav/Downloads/eclipse/p2/org.eclipse.equinox.p2.repository/cache/content475293631.jar

Interestingly enough, I haven't found a content475293631.jar on the
jenkins site and content.jar has a different md5sum and size when
downloaded with a browser.

the complete download of
geppetto-win32.win32.x86_64-4.2.0-I201407160516.zip worked and was able
to load my project.

Regards, David

On 2014-07-24 01:23, Thomas Hallgren wrote:
> Hi Brian,
>
> I'm using both the new URL (space removed):
>
>
>
> http://geppetto-jenkins.puppetlabs.com/job/GeppettoDev/lastSuccessfulBuild/artifact/geppetto.p2.repository/

> [4]
>
> and the old one:
>
> http://geppetto-jenkins.puppetlabs.com/job/Geppetto [5]

>> in JENKINS-19752 [1].  However, I have noticed that several others

>> have been able to use the p2 site without issue (once the space was
>> addressed).  Perhaps Jenkins was recently updated to a version that
>> includes this defect?
>>
>> Thanks,
>> Brian  --
>> You received this message because you are subscribed to the Google
>> Groups "Puppet Geppetto" group.
>> To unsubscribe from this group and stop receiving emails from it,

>> send an email to puppet-geppet...@googlegroups.com [2].
>> For more options, visit https://groups.google.com/d/optout [3].

>
> --
> You received this message because you are subscribed to the Google
> Groups "Puppet Geppetto" group.
> To unsubscribe from this group and stop receiving emails from it,

> send an email to puppet-geppet...@googlegroups.com [6].
> For more options, visit https://groups.google.com/d/optout [7].
>
>
> Links:
> ------
> [1] https://issues.jenkins-ci.org/browse/JENKINS-19752
> [2] mailto:puppet-geppet...@googlegroups.com
> [3] https://groups.google.com/d/optout
> [4]
>
> http://geppetto-jenkins.puppetlabs.com/job/GeppettoDev/lastSuccessfulBuild/artifact/geppetto.p2.repository/
> [5] http://geppetto-jenkins.puppetlabs.com/job/Geppetto
> [6] mailto:puppet-geppet...@googlegroups.com
> [7] https://groups.google.com/d/optout

[David Schmitt]

Hi,

some notes on what I noticed with the new version. Since I'm currently
not in my office, only an informal listing. I can create bugs later if
you want.


* "Must be a name or type (all segments must start with same
case). connector.pp /tomcat/manifests line: 94
/tomcat/manifests/connector.pp PP Problem"
caused by variable starting with an underscore.

* double clicking on the error opens the file with cursor in the
first line; needs second double click after opening to jump to correct
line

* Modulefiles seem to have lost their special nature: Default editor
opens the file as text, icon indicates text too.

* Default encoding and line endings are cp1252 and windows. Default
for interoperability should be utf-8 and unix.

* "Unknown resource type: 'package' FOO.pp /FOO/manifests line: 35
/FOO/manifests/FOO.pp PP Problem"
"Unknown attribute: 'ensure' in definition:
'logrotate.file' init.pp /logrotate/manifests line: 9
/logrotate/manifests/init.pp PP Problem"
"Unknown function: 'defined' service.pp /apache/manifests line: 25
/apache/manifests/service.pp PP Problem"

On build (new workspace) these errors are reported for all native
resources types, functions and parameters.

* Creating a new module only helps partially: regular types are now
recognized, but instead of a Modulefile, a metadata.json file is
created. The latter is editable only as text, but provides syntax
highlighting and error reporting. Looking into the generated .project
file, compared to my usual ones, it has removed the
com.puppetlabs.geppetto.pp.dsl.ui.modulefileBuilder builder and no other
changes.

* Using the native proxy provider didn't work. Configuring the proxy
manually worked. I cannot guarantee that that's not a local problem and
it already appeared with an empty eclipse 4.4.


Enough for now, I'm stepping back to 4.1 as 4.2 is currently not usable
for me.

Regards, David

[Thomas Hallgren]

Hi David,

Thanks for trying out the new version giving feedback. I think some of the issues that you see is actually expected
behavior and a two others are issues that have been reported already. More comments below.

On 2014-07-24 10:17, David Schmitt wrote:
> Hi,
>
> some notes on what I noticed with the new version. Since I'm currently not in my office, only an informal listing. I
> can create bugs later if you want.

Yes, please do.

>
>
> * "Must be a name or type (all segments must start with same case). connector.pp /tomcat/manifests line: 94
> /tomcat/manifests/connector.pp PP Problem"
> caused by variable starting with an underscore.

Is this the ${_basedir} in camptocamp-tomcat? In that case it's a known bug and you should see it in 4.1 as well.

https://tickets.puppetlabs.com/browse/GEP-120

>
> * double clicking on the error opens the file with cursor in the first line; needs second double click after opening
> to jump to correct line
>

That seems like a bug although I'm not able to reproduce it. Please enter a JIRA ticket for this one and if possible,
attach something that makes it easy to reproduce.

> * Modulefiles seem to have lost their special nature: Default editor opens the file as text, icon indicates text too.

This is intentional. The Modulefile is deprecated and should no longer be used. The metadata.json is now the primary
source of metadata for a module.

>
> * Default encoding and line endings are cp1252 and windows. Default for interoperability should be utf-8 and unix.
>

I agree. This bug has been reported:

https://tickets.puppetlabs.com/browse/GEP-267

I just didn't have time to fix it in time for 4.2.

> * "Unknown resource type: 'package' FOO.pp /FOO/manifests line: 35 /FOO/manifests/FOO.pp PP Problem"
> "Unknown attribute: 'ensure' in definition: 'logrotate.file' init.pp /logrotate/manifests line: 9
> /logrotate/manifests/init.pp PP Problem"
> "Unknown function: 'defined' service.pp /apache/manifests line: 25 /apache/manifests/service.pp PP Problem"
>
> On build (new workspace) these errors are reported for all native resources types, functions and parameters.
>

I'm not able to reproduce this. Would it be possible for you to attach a workspace that exhibits this problem to a JIRA?

> * Creating a new module only helps partially: regular types are now recognized, but instead of a Modulefile, a
> metadata.json file is created. The latter is editable only as text, but provides syntax highlighting and error
> reporting. Looking into the generated .project file, compared to my usual ones, it has removed the
> com.puppetlabs.geppetto.pp.dsl.ui.modulefileBuilder builder and no other changes.

These changes are intentional and one of the major changes in 4.2. The reason the builder is removed is that it no
longer serves any purpose. It's been replaced by the same type of validation logic that is used when validating puppet
manifests. This is also what provides the tight integration between the validator and the text based editor.

>
> * Using the native proxy provider didn't work. Configuring the proxy manually worked. I cannot guarantee that that's
> not a local problem and it already appeared with an empty eclipse 4.4.
>

- thomas

[Henrik Lindberg]

Thanks for testing and reporting - some comments inline below.

- henrik

On 2014-24-07 10:17, David Schmitt wrote:
> Hi,
>
> some notes on what I noticed with the new version. Since I'm currently
> not in my office, only an informal listing. I can create bugs later if
> you want.
>
>
> * "Must be a name or type (all segments must start with same
> case). connector.pp /tomcat/manifests line: 94
> /tomcat/manifests/connector.pp PP Problem"
> caused by variable starting with an underscore.
>

Which target platform? Future? Can you paste snippet of the puppet
construct that causes this.

> * double clicking on the error opens the file with cursor in the
> first line; needs second double click after opening to jump to correct line
>

Bug.

> * Modulefiles seem to have lost their special nature: Default editor
> opens the file as text, icon indicates text too.
>

Thomas needs to comment on this.

> * Default encoding and line endings are cp1252 and windows. Default
> for interoperability should be utf-8 and unix.
>

Yeah, the default follows the platform default. We should consider
having it hardcoded in the packaged Geppetto.

> * "Unknown resource type: 'package' FOO.pp /FOO/manifests
> line: 35 /FOO/manifests/FOO.pp PP Problem"
> "Unknown attribute: 'ensure' in definition: 'logrotate.file'
> init.pp /logrotate/manifests line: 9
> /logrotate/manifests/init.pp PP Problem"
> "Unknown function: 'defined' service.pp /apache/manifests
> line: 25 /apache/manifests/service.pp PP Problem"
>
> On build (new workspace) these errors are reported for all native
> resources types, functions and parameters.
>

Bug.
Does a Build Clean fix it?

> * Creating a new module only helps partially: regular types are now
> recognized, but instead of a Modulefile, a metadata.json file is
> created. The latter is editable only as text, but provides syntax
> highlighting and error reporting. Looking into the generated .project
> file, compared to my usual ones, it has removed the
> com.puppetlabs.geppetto.pp.dsl.ui.modulefileBuilder builder and no other
> changes.
>

Sounds like a bug.

>>>> puppet-geppetto+unsubscribe-/JYPxA39Uh5...@public.gmane.org

>>>> [2].
>>>> For more options, visit https://groups.google.com/d/optout [3].
>>>
>>> --
>>> You received this message because you are subscribed to the Google
>>> Groups "Puppet Geppetto" group.
>>> To unsubscribe from this group and stop receiving emails from it,
>>> send an email to

>>> puppet-geppetto+unsubscribe-/JYPxA39Uh5...@public.gmane.org [6].

>>> For more options, visit https://groups.google.com/d/optout [7].
>>>
>>>
>>> Links:
>>> ------
>>> [1] https://issues.jenkins-ci.org/browse/JENKINS-19752
>>> [2]

>>> mailto:puppet-geppetto+unsubscribe-/JYPxA39Uh5...@public.gmane.org

>>>
>>> [3] https://groups.google.com/d/optout
>>> [4]
>>>
>>> http://geppetto-jenkins.puppetlabs.com/job/GeppettoDev/lastSuccessfulBuild/artifact/geppetto.p2.repository/
>>>
>>> [5] http://geppetto-jenkins.puppetlabs.com/job/Geppetto
>>> [6]

>>> mailto:puppet-geppetto+unsubscribe-/JYPxA39Uh5...@public.gmane.org
>>>
>>> [7] https://groups.google.com/d/optout
>


--

Visit my Blog "Puppet on the Edge"
http://puppet-on-the-edge.blogspot.se/

[Thomas Hallgren]

On 2014-07-24 09:48, David Schmitt wrote:

> Hi,
>
> I've got the same error:
>
>> Unable to read repository at
>> file:/C:/Users/schmidav/Downloads/eclipse/p2/org.eclipse.equinox.p2.repository/cache/content475293631.jar.
>> Unable to read repository at
>> file:/C:/Users/schmidav/Downloads/eclipse/p2/org.eclipse.equinox.p2.repository/cache/content475293631.jar.
>> invalid distance too far back
>
> This is a fresh eclipse 4.4 install on windows 7.

I obviously need to test this thoroughly on a Windows box. I have no idea what's going on here. Can you please add a
JIRA ticket for this?

- thomas

[Thomas Hallgren]

Regarding the proxy problem:

On 2014-07-24 10:17, David Schmitt wrote:
>
>
> * Using the native proxy provider didn't work. Configuring the proxy manually worked. I cannot guarantee that that's
> not a local problem and it already appeared with an empty eclipse 4.4.
>

Since you run under windows, I strongly suspect that what you see is what's described here:

https://bugs.eclipse.org/bugs/show_bug.cgi?id=422665

I'll try to fix this for Geppetto.

- thomas

[Brian Keyser]

Hi Thomas,

I have added an incident in the Puppet issue tracking system currently identified as SCM-7 describing the problem that occurs on both Windows 7 and CentOS 6.5.

Thanks for your efforts,

Brian

--
You received this message because you are subscribed to a topic in the Google Groups "Puppet Geppetto" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/puppet-geppetto/5XpfrJUQlqw/unsubscribe.
To unsubscribe from this group and all its topics, send an email to puppet-geppetto+unsubscribe@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Kenn Hussey]

Thanks for submitting the issue, Brian. It's been moved to GEP-153.

Kenn

--
You received this message because you are subscribed to the Google Groups "Puppet Geppetto" group.
To unsubscribe from this group and stop receiving emails from it, send an email to puppet-geppet...@googlegroups.com.

[David Schmitt]

On 2014-07-24 14:58, Thomas Hallgren wrote:
> Hi David,
>
> Thanks for trying out the new version giving feedback. I think some of
> the issues that you see is actually expected behavior and a two others
> are issues that have been reported already. More comments below.
>
> On 2014-07-24 10:17, David Schmitt wrote:
>> Hi,
>>
>> some notes on what I noticed with the new version. Since I'm currently
>> not in my office, only an informal listing. I can create bugs later if
>> you want.
> Yes, please do.
>
>>
>>
>> * "Must be a name or type (all segments must start with same
>> case). connector.pp /tomcat/manifests line: 94
>> /tomcat/manifests/connector.pp PP Problem"
>> caused by variable starting with an underscore.
>
> Is this the ${_basedir} in camptocamp-tomcat?

How did you know? ;-)

> In that case it's a known
> bug and you should see it in 4.1 as well.
>
> https://tickets.puppetlabs.com/browse/GEP-120

The power of looking the other direction. Yes, I do see that in 4.1 too.
(On this particular project, there are "still" 70+ errors with 4.1).

>> * double clicking on the error opens the file with cursor in the
>> first line; needs second double click after opening to jump to correct
>> line
>>
> That seems like a bug although I'm not able to reproduce it. Please
> enter a JIRA ticket for this one and if possible, attach something that
> makes it easy to reproduce.

I can't repro that on Linux now. The other project is not sharable, sorry.

>> * Modulefiles seem to have lost their special nature: Default editor
>> opens the file as text, icon indicates text too.
>
> This is intentional. The Modulefile is deprecated and should no longer
> be used. The metadata.json is now the primary source of metadata for a
> module.

Is there any migration available/possible? I've later noted, that 4.1
can already open metadata.json with a rich editor, which is not used in
4.2 for metadata.json ...

>> * Default encoding and line endings are cp1252 and windows. Default
>> for interoperability should be utf-8 and unix.
>>
> I agree. This bug has been reported:
>
> https://tickets.puppetlabs.com/browse/GEP-267
>
> I just didn't have time to fix it in time for 4.2.

n.p., just a nuisance.

>> * "Unknown resource type: 'package' FOO.pp /FOO/manifests
>> line: 35 /FOO/manifests/FOO.pp PP Problem"
>> "Unknown attribute: 'ensure' in definition: 'logrotate.file'
>> init.pp /logrotate/manifests line: 9
>> /logrotate/manifests/init.pp PP Problem"
>> "Unknown function: 'defined' service.pp /apache/manifests
>> line: 25 /apache/manifests/service.pp PP Problem"
>>
>> On build (new workspace) these errors are reported for all native
>> resources types, functions and parameters.
>>
> I'm not able to reproduce this. Would it be possible for you to attach a
> workspace that exhibits this problem to a JIRA?

https://tickets.puppetlabs.com/browse/GEP-154

>> * Creating a new module only helps partially: regular types are now
>> recognized, but instead of a Modulefile, a metadata.json file is
>> created. The latter is editable only as text, but provides syntax
>> highlighting and error reporting. Looking into the generated .project
>> file, compared to my usual ones, it has removed the
>> com.puppetlabs.geppetto.pp.dsl.ui.modulefileBuilder builder and no
>> other changes.
>
> These changes are intentional and one of the major changes in 4.2. The
> reason the builder is removed is that it no longer serves any purpose.
> It's been replaced by the same type of validation logic that is used
> when validating puppet manifests. This is also what provides the tight
> integration between the validator and the text based editor.

As written above: migration path or hint in the release notes would be nice

And the rich editor is not started.

>> * Using the native proxy provider didn't work. Configuring the proxy
>> manually worked. I cannot guarantee that that's not a local problem
>> and it already appeared with an empty eclipse 4.4.
>>

> Since you run under windows, I strongly suspect that what you see is what's described here:
>
> https://bugs.eclipse.org/bugs/show_bug.cgi?id=422665

O.o

> I'll try to fix this for Geppetto.

like the default encoding it's just a nuisance that can be fixed by
educating users.


Regards, David
--
* Always looking for people I can help with awesome projects *
G+: https://plus.google.com/+DavidSchmitt
Blog: http://club.black.co.at/log/
LinkedIn: http://at.linkedin.com/in/davidschmitt

[Thomas Hallgren]

On 2014-07-24 22:17, David Schmitt wrote:

I'm not able to reproduce this. Would it be possible for you to attach a
workspace that exhibits this problem to a JIRA?

https://tickets.puppetlabs.com/browse/GEP-154

Thanks.

  * Creating a new module only helps partially: regular types are now
recognized, but instead of a Modulefile, a metadata.json file is
created. The latter is editable only as text, but provides syntax
highlighting and error reporting. Looking into the generated .project
file, compared to my usual ones, it has removed the
com.puppetlabs.geppetto.pp.dsl.ui.modulefileBuilder builder and no
other changes.

These changes are intentional and one of the major changes in 4.2. The
reason the builder is removed is that it no longer serves any purpose.
It's been replaced by the same type of validation logic that is used
when validating puppet manifests. This is also what provides the tight
integration between the validator and the text based editor.

As written above: migration path or hint in the release notes would be nice

And the rich editor is not started.

The "migration" that Geppetto offers consists of the following:

• The user is notified that the Modulefile is deprecated (configurable setting).
• If metadata.json is missing it is created from the Modulefile if it exists.
• Directories that only contain a Modulefile are still recognized as module roots.
• A release note for 4.2 is in the making that explains the change in detail. It will accompany the official release.
  

The old forms based editor was removed. There were several reasons for this:

Keeping it means maintaining it and thus add forms based editing of all new metadata.json fields (now and in the future). Those fields have no representation in the Modulefile so that part of the editor would be broken anyway.

It would also mean maintaining large parts of the underlying parsing/validation structure that is now managed by the xtext infrastructure that the new editor is built on top of.

The new text-based editor is, from a feature perspective, richer than the old one. There, you didn't find a documentation tooltip for all keywords. There was no proposal of valid version constructs in a dependency. No quick fix that makes it easy to import a missing dependency. No automatic linking that allowed you to jump from a dependency to the module it references. No configurable preference settings.

- thomas

[David Schmitt]

On 25.07.2014 07:43, Thomas Hallgren wrote:
> On 2014-07-24 22:17, David Schmitt wrote:
>>>> * Creating a new module only helps partially: regular types are now
>>>> recognized, but instead of a Modulefile, a metadata.json file is
>>>> created. The latter is editable only as text, but provides syntax
>>>> highlighting and error reporting. Looking into the generated .project
>>>> file, compared to my usual ones, it has removed the
>>>> com.puppetlabs.geppetto.pp.dsl.ui.modulefileBuilder builder and no
>>>> other changes.
>>>
>>> These changes are intentional and one of the major changes in 4.2. The
>>> reason the builder is removed is that it no longer serves any purpose.
>>> It's been replaced by the same type of validation logic that is used
>>> when validating puppet manifests. This is also what provides the tight
>>> integration between the validator and the text based editor.
>>
>> As written above: migration path or hint in the release notes would be
>> nice
>>
>> And the rich editor is not started.
>>
> The "migration" that Geppetto offers consists of the following:
>

> * The user is notified that the Modulefile is deprecated (configurable
> setting).
> * If metadata.json is missing it is created from the Modulefile if it
> exists.

Ah, I haven't noticed those two. Must look harder.

> * Directories that only contain a Modulefile are still recognized as
> module roots.
> * A release note for 4.2 is in the making that explains the change in

> detail. It will accompany the official release.

... may I suggest adding draft release notes to the next testing
announcement?

> The old forms based editor was removed. There were several reasons for this:

[good reasons snipped]

point taken :-)


Regards, David

[David Schmitt]

On 24.07.2014 14:58, Thomas Hallgren wrote:
>> * "Unknown resource type: 'package' FOO.pp /FOO/manifests
>> line: 35 /FOO/manifests/FOO.pp PP Problem"
>> "Unknown attribute: 'ensure' in definition: 'logrotate.file'
>> init.pp /logrotate/manifests line: 9
>> /logrotate/manifests/init.pp PP Problem"
>> "Unknown function: 'defined' service.pp /apache/manifests
>> line: 25 /apache/manifests/service.pp PP Problem"
>>
>> On build (new workspace) these errors are reported for all native
>> resources types, functions and parameters.
>>
> I'm not able to reproduce this. Would it be possible for you to attach a
> workspace that exhibits this problem to a JIRA?

Hmm, I've just repro'd this with 4.1 and a different workspace at
another customer. Again, nothing I can share, but at "least" it doesn't
seem to be a regression in 4.2. Although I cannot exclude the
possibility that the running 4.2 on this machine hasn't infected my
profile somehow.


Regards, David

[Thomas Hallgren]

And you still get it with 4.2?

- thomas