Home » Archived » M2M (model-to-model transformation) » Headless configurations
Headless configurations [message #103131] |
Wed, 08 April 2009 22:16 |
|
Hi,
I'm trying to use QVT in a headless environment. One requirement is that we keep it mean and lean without UI
dependencies. It turned out to be difficult due to these hard dependencies
org.eclipse.m2m.qvt.oml.common => org.eclipse.ui
org.eclipse.m2m.qvt.oml.runtime => org.eclipse.ui.externaltools
From the looks of it, there's a very good separation of concern between UI and non-UI components throughout, so perhaps
this is subject to change? From what I can see, at least the dependency in the runtime bundle is redundant.
Kind Regards,
Thomas Hallgren
|
|
|
Re: Headless configurations [message #103153 is a reply to message #103131] |
Thu, 09 April 2009 09:07 |
Radomil Dvorak Messages: 249 Registered: July 2009 |
Senior Member |
|
|
Hi Thomas,
Thanks for catching this!
I will fix that for tomorrow QVTo I-build, as I guess you are affected
in "[Bug 270058] [releng] Galileo build and process needs to be improved".
Regards,
/Radek
On Thu, 09 Apr 2009 00:16:12 +0200, Thomas Hallgren <thomas@tada.se> wrote:
> Hi,
> I'm trying to use QVT in a headless environment. One requirement is that
> we keep it mean and lean without UI dependencies. It turned out to be
> difficult due to these hard dependencies
>
> org.eclipse.m2m.qvt.oml.common => org.eclipse.ui
> org.eclipse.m2m.qvt.oml.runtime => org.eclipse.ui.externaltools
>
> From the looks of it, there's a very good separation of concern between
> UI and non-UI components throughout, so perhaps this is subject to
> change? From what I can see, at least the dependency in the runtime
> bundle is redundant.
>
> Kind Regards,
> Thomas Hallgren
|
|
|
Re: Headless configurations [message #103199 is a reply to message #103153] |
Thu, 09 April 2009 11:06 |
|
Hi Radek,
Radek Dvorak wrote:
> I will fix that for tomorrow QVTo I-build, as I guess you are affected
> in " [releng] Galileo build and process needs to be improved".
>
Yes, that's the primary use-case :-), but I'm also looking into integrating Buckminster a lot more with modeling when
going forward.
I found two other things that would make life easier for a headless consumer. They are not the same magnitude as the UI
dependency but I think they might be worth mentioning:
1. org.eclipse.ocl has a package dependency to com.ibm.icu.lang;version="[4.0.0,5.0.0)"
This is not ideal since this package is excluded from com.ibm.icu.base so in essence, this forces a dependency to the
com.ibm.icu bundle. The reason it's there is that the UCharacter class is used for toUpper and toLower translations. It
can be argued that the standard java.lang.Character and java.lang.String actually does a better job. At least in more
modern JVM's (discussion here: http://dev.eclipse.org/mhonarc/lists/cross-project-issues-de v/msg02975.html).
2. org.eclipse.ocl.ecore has an optional dependency to com.ibm.icu [3.4.0,4.0.0)
This dependency does not resolve without a conflict since other packages will require >= 4.0.0. Question is, why is it
there in the first place? Nothing bad seems to happen if I remove it.
Regards,
Thomas Hallgren
|
|
|
Re: Headless configurations [message #103708 is a reply to message #103153] |
Fri, 17 April 2009 08:49 |
|
Hi Radek,
Any news on this? I'm a bit uncertain where you publish your I-builds.
Regards,
Thomas Hallgren
Radek Dvorak wrote:
> Hi Thomas,
>
> Thanks for catching this!
>
> I will fix that for tomorrow QVTo I-build, as I guess you are affected
> in "[Bug 270058] [releng] Galileo build and process needs to be improved".
>
> Regards,
> /Radek
>
>
> On Thu, 09 Apr 2009 00:16:12 +0200, Thomas Hallgren <thomas@tada.se> wrote:
>
>> Hi,
>> I'm trying to use QVT in a headless environment. One requirement is
>> that we keep it mean and lean without UI dependencies. It turned out
>> to be difficult due to these hard dependencies
>>
>> org.eclipse.m2m.qvt.oml.common => org.eclipse.ui
>> org.eclipse.m2m.qvt.oml.runtime => org.eclipse.ui.externaltools
>>
>> From the looks of it, there's a very good separation of concern
>> between UI and non-UI components throughout, so perhaps this is
>> subject to change? From what I can see, at least the dependency in the
>> runtime bundle is redundant.
>>
>> Kind Regards,
>> Thomas Hallgren
>
|
|
|
Re: Headless configurations [message #103728 is a reply to message #103708] |
Fri, 17 April 2009 10:05 |
Radomil Dvorak Messages: 249 Registered: July 2009 |
Senior Member |
|
|
Hi Thomas,
Actually, I made that I-build on the last Fri, but now I can see it has
not been published.
Though, I was convinced I had fired promote (did not verify the outcome
this time).
Or perhaps, I just had a releng 'fata morgana' on that late Friday evening.
I have promoted that I-build now, it should come out in a while.
As for I-build publishing, we had more frequent I-build occurences before
EclipseCon and a small gap right after as nothing new actually happened.
Normally, it should be in the end of the week, before the GMF I-build as
its Xpand
implementation has developmnet depencencies on QVTo.
I can see the build has just arrived
http://www.eclipse.org/modeling/m2m/downloads/index.php?show All=1&hlbuild=I200904101313&project=qvtoml#I20090410 1313
Regards,
/Radek
On Fri, 17 Apr 2009 10:49:47 +0200, Thomas Hallgren <thomas@tada.se> wrote:
> Hi Radek,
> Any news on this? I'm a bit uncertain where you publish your I-builds.
>
> Regards,
> Thomas Hallgren
>
>
> Radek Dvorak wrote:
>> Hi Thomas,
>> Thanks for catching this!
>> I will fix that for tomorrow QVTo I-build, as I guess you are affected
>> in "[Bug 270058] [releng] Galileo build and process needs to be
>> improved".
>> Regards,
>> /Radek
>> On Thu, 09 Apr 2009 00:16:12 +0200, Thomas Hallgren <thomas@tada.se>
>> wrote:
>>
>>> Hi,
>>> I'm trying to use QVT in a headless environment. One requirement is
>>> that we keep it mean and lean without UI dependencies. It turned out
>>> to be difficult due to these hard dependencies
>>>
>>> org.eclipse.m2m.qvt.oml.common => org.eclipse.ui
>>> org.eclipse.m2m.qvt.oml.runtime => org.eclipse.ui.externaltools
>>>
>>> From the looks of it, there's a very good separation of concern
>>> between UI and non-UI components throughout, so perhaps this is
>>> subject to change? From what I can see, at least the dependency in the
>>> runtime bundle is redundant.
>>>
>>> Kind Regards,
>>> Thomas Hallgren
>>
|
|
|
Goto Forum:
Current Time: Sun Apr 28 03:00:25 GMT 2024
Powered by FUDForum. Page generated in 0.04078 seconds
|