Eclipse Community Forums
Forum Search:

Search      Help    Register    Login    Home
Home » Modeling » QVT-Relations » Current status of QVTd?
Current status of QVTd? [message #1006608] Fri, 01 February 2013 02:16 Go to next message
Akira Tanaka is currently offline Akira Tanaka
Messages: 31
Registered: March 2010
Member
Hello QVT-Declarative experts,

I am curious about the current status and future plan of QVT-Declarative.

I tried the following.

- Installed eclipse modeling (Juno) package
- Installed QVT Declarative 0.9.0 using All-In-One Update site [also tried other versions but did not work]
- Installed OCL package using "Install Modeling Components" (to utilize ModelRegistry Model Editor)
- Introduced ModelMorf distribution Examples into the workspace

I see introduced *.qvtr files have lots of red flags. It seems most of them are due to the fact that they are unable to recognize metamodels (.emof files -- even modified as ecore files). I am curious to learn if there is any way to fix this. If not, is there any plan?

It seems QVT Declarative is not usable at this moment (if above is not fixable). Or, is there any configurations or previous package etc. we can use for the time being? Also, with the installation of QVT Declarative, it provides (from File -> New menu) creating empty QVTc project capability. Is QVTc actually implemented in the same distribution?

Thank you for clarification in advance.

Akira
Re: Current status of QVTd? [message #1006613 is a reply to message #1006608] Fri, 01 February 2013 02:39 Go to previous messageGo to next message
Ed Willink is currently offline Ed Willink
Messages: 4027
Registered: July 2009
Senior Member
Hi

The QVTc and QVTr editors should be usefully functional. The 0.10
editors from the download pages may be slightly better.

The ModelMorf examples had only a couple of errors with the 0.8
IMP-based editors. I must confess that I haven't checked them properly
with the newer Xtext-based editors. I suspect that they just need import
declarations. In 0.8, a Model Registry was required to specify the
meta-model name to location mapping. I eventually decided that this was
a language bug and introduced import, include and library declarations.
Look at the Rel2Core which is what I use mostly for exercising.

My company, Willink Transformations, is sponsoring an EngD student at
York to pursue the progressive transformation approach outlined in
http://wiki.eclipse.org/MMT/QVT_Declarative_Languages. Currently some
very simple QVTi transformations are working. This gives confidence that
the approach is good. The execution work is all Open Source and
currently on GitHub at git://github.com/arcanefoam/qvtMustus. It will
move to Eclipse in a few months.

The goal is basic QVTc execution for Kepler, basic QVTr execution for
Kepler+1, optimized execution for Kepler+2.

The code is all layered on top of the new Xtext-based, UML-aligned pivot
OCL binding, so the new direct OCL to Java code generator can be
exploited to give high quality compiled transformation code.

Regards

Ed Willink



On 01/02/2013 07:16, Akira Tanaka wrote:
> Hello QVT-Declarative experts,
>
> I am curious about the current status and future plan of QVT-Declarative.
> I tried the following.
>
> - Installed eclipse modeling (Juno) package
> - Installed QVT Declarative 0.9.0 using All-In-One Update site [also
> tried other versions but did not work]
> - Installed OCL package using "Install Modeling Components" (to
> utilize ModelRegistry Model Editor)
> - Introduced ModelMorf distribution Examples into the workspace
>
> I see introduced *.qvtr files have lots of red flags. It seems most of
> them are due to the fact that they are unable to recognize metamodels
> (.emof files -- even modified as ecore files). I am curious to learn
> if there is any way to fix this. If not, is there any plan?
>
> It seems QVT Declarative is not usable at this moment (if above is not
> fixable). Or, is there any configurations or previous package etc. we
> can use for the time being? Also, with the installation of QVT
> Declarative, it provides (from File -> New menu) creating empty QVTc
> project capability. Is QVTc actually implemented in the same
> distribution?
>
> Thank you for clarification in advance.
>
> Akira
Re: Current status of QVTd? [message #1006809 is a reply to message #1006613] Sat, 02 February 2013 01:11 Go to previous messageGo to next message
Akira Tanaka is currently offline Akira Tanaka
Messages: 31
Registered: March 2010
Member
Hello,

Thank you for the explanation. Maybe I should wait for Kepler in June.

With your advice, I did the following (with the results).

1) "import" statement
With 0.9.0, the statements of the following styles did not work.

- style #1: "import SimpleUML : 'SimpleUml.ecore'::umlMM;"
- style #2: "import SimpleUML : 'SimpleUml.ecore#/'"

For instance, if I take simpler example "FlatRelToCore.qvtr"

when I did the following (after converting emof to ecore):

import QVTMM : 'FlatQVT.ecore'::FlatQVT;


"transformation" statement gets red flag saying "mismatched input '::' expecting ';'"

and if I did the following

import QVTMM : 'FlatQVT.ecore#/';


"import" statement gets red flag saying "mismatched input '::' expecting ';'"

I looked at the github sample code you mentioned and found 0.10.0 is used instead of 0.9.0. So I uninstalled 0.9.0 and tried to install 0.10.0M4, but could not.

Cannot complete the install because one or more required items could not be found.
  Software being installed: QVT Declarative - All examples 0.10.0.v20121103-1009 (org.eclipse.qvtd.examples.feature.group 0.10.0.v20121103-1009)
  Missing requirement: QVT Declarative Examples 0.10.0.v20121103-1009 (org.eclipse.qvtd.examples 0.10.0.v20121103-1009) requires 'bundle org.eclipse.core.runtime [3.9.0,4.0.0)' but it could not be found
  Cannot satisfy dependency:
    From: QVT Declarative - All examples 0.10.0.v20121103-1009 (org.eclipse.qvtd.examples.feature.group 0.10.0.v20121103-1009)
    To: org.eclipse.qvtd.examples [0.10.0.v20121103-1009]


I tried above with standard Juno modeling package. Do I need to use early version of Kepler?

Thank you for more help in advance.

Akira
Re: Current status of QVTd? [message #1006813 is a reply to message #1006809] Sat, 02 February 2013 05:55 Go to previous messageGo to next message
Ed Willink is currently offline Ed Willink
Messages: 4027
Registered: July 2009
Senior Member
Hi

There is a version of RelToCore in the JUnit tests plugin where many
more errors detected by better OCL tooling have been fixed. It seems I
forgot to migrate this back to the examples.

On 02/02/2013 06:11, Akira Tanaka wrote:
> Thank you for the explanation. Maybe I should wait for Kepler in June.
I'll try to make the examples much more coherent for M5 (8th Feb).
> I looked at the github sample code you mentioned and found 0.10.0 is
> used instead of 0.9.0. So I uninstalled 0.9.0 and tried to install
> 0.10.0M4, but could not.
>
> Cannot satisfy dependency:
> From: QVT Declarative - All examples 0.10.0.v20121103-1009
> (org.eclipse.qvtd.examples.feature.group 0.10.0.v20121103-1009)
> To: org.eclipse.qvtd.examples [0.10.0.v20121103-1009]
Thank you for pointing this out. I've fixed the missing bounds that the
builder converts to current bounds. the N201302020523 build (and so
Kepler M5) is now installable on Juno (but you will also need OCL M5 and
at least Xtext M1 which may most easily be found in
http://download.eclipse.org/releases/staging.)

Regards

Ed Willink
Re: Current status of QVTd? [message #1006854 is a reply to message #1006813] Sat, 02 February 2013 17:59 Go to previous messageGo to next message
Ed Willink is currently offline Ed Willink
Messages: 4027
Registered: July 2009
Senior Member
Hi

On 02/02/2013 10:55, Ed Willink wrote:
> There is a version of RelToCore in the JUnit tests plugin where many
> more errors detected by better OCL tooling have been fixed. It seems I
> forgot to migrate this back to the examples.
For M5, parsing of the examples is now JUnit-tested. RelToCore and seven
out of eleven of the ModelMorf examples are error-free.

Regards

Ed Willink
Re: Current status of QVTd? [message #1006860 is a reply to message #1006854] Sat, 02 February 2013 19:59 Go to previous messageGo to next message
Akira Tanaka is currently offline Akira Tanaka
Messages: 31
Registered: March 2010
Member
Thank you for the update.

I will be waiting for the N201302020523 qvtd build showing up on "http://www.eclipse.org/mmt/downloads/?project=qvtd" page. Also,

Is OCL 4.1.0.v2013.02.01 enough to cover what you call "OCL M5"? And, is Xtext SDK 2.4.0.v20130201105 enough to cover what you call "Xtext M1"? [I think both are the latest.]

Thank you for your help again.

Akira
Re: Current status of QVTd? [message #1006969 is a reply to message #1006860] Sun, 03 February 2013 06:13 Go to previous messageGo to next message
Ed Willink is currently offline Ed Willink
Messages: 4027
Registered: July 2009
Senior Member
Hi

The promotion from the build pages to the download pages is polled; once
a day for OCL.

The required OCL is now there:
http://www.eclipse.org/modeling/download.php?file=/modeling/mdt/ocl/downloads/drops/4.1.0/I201302021655/mdt-ocl-Update-tools-I201302021655.zip

The required QVTd is there too:
http://www.eclipse.org/downloads/download.php?file=/mmt/qvtd/downloads/drops/0.10.0/N201302021656/mmt-qvtd-incubation-Update-N201302021656.zip

Xtext SDK 2.4.0.v20130201105 should be more than enough; just need 2.4
rather than 2.3 to satisfy lower bounds.

Regards

Ed Willink

On 03/02/2013 00:59, Akira Tanaka wrote:
> Thank you for the update.
>
> I will be waiting for the N201302020523 qvtd build showing up on
> "http://www.eclipse.org/mmt/downloads/?project=qvtd" page. Also,
>
> Is OCL 4.1.0.v2013.02.01 enough to cover what you call "OCL M5"? And,
> is Xtext SDK 2.4.0.v20130201105 enough to cover what you call "Xtext
> M1"? [I think both are the latest.]
>
> Thank you for your help again.
>
> Akira
Re: Current status of QVTd? [message #1007052 is a reply to message #1006969] Mon, 04 February 2013 04:23 Go to previous messageGo to next message
Akira Tanaka is currently offline Akira Tanaka
Messages: 31
Registered: March 2010
Member
Hello,

FInally I can use the QVTd editor! Thank you very much for your help and efforts.

Best Regards,
Akira
Re: Current status of QVTd? [message #1069135 is a reply to message #1007052] Mon, 15 July 2013 04:20 Go to previous messageGo to next message
Goood Guy is currently offline Goood Guy
Messages: 126
Registered: September 2011
Senior Member
Hi,

just recently I became really aware of the QVTd project, which is very exciting. Since 0.10.0 was recently released, I am interessted in an update of the project status. Especially if the release already comes with an execution implementation and how the execution can be triggered.


Simon
Re: Current status of QVTd? [message #1069160 is a reply to message #1069135] Mon, 15 July 2013 05:43 Go to previous message
Ed Willink is currently offline Ed Willink
Messages: 4027
Registered: July 2009
Senior Member
Hi

There is an interpreted and a code generated execution capability for
QVTi inherited from OCL. (Probably the first M2M project with Java code
generation at Eclipse).

However QVTi is just a simple imperative semantics. Just like assembler,
very useful to have, but you wouldn't want to use it manually.

We hope to get M2M transformation chain from QVTr to QVTc to QVTu to
QVTm to QVTi in place for Luna

See wiki.eclipse.org/MMT/QVT_Declarative_Languages,
http://www.slideshare.net/EdWillink/icmt2013-talk,
http://link.springer.com/chapter/10.1007%2F978-3-642-38883-5_8

We also hope to have a debugger for Luna.

Advanced optimisations for QVTr are targetted for Luna+1 and Luna+2.

So we now have editors/parsers and low level execution. Just need to
transform across the abstraction gap from QVTr/c to QVTi.

Regards

Ed Willink


On 15/07/2013 09:21, Goood Guy wrote:
> Hi,
>
> just recently I became really aware of the QVTd project, which is very
> exciting. Since 0.10.0 was recently released, I am interessted in an
> update of the project status. Especially if the release already comes
> with an execution implementation and how the execution can be triggered.
>
>
> Simon
Previous Topic:QVT expressivity
Next Topic:Primitive Domains in Top Relations
Goto Forum:
  


Current Time: Fri Aug 22 21:51:05 EDT 2014

Powered by FUDForum. Page generated in 0.10297 seconds