Skip to main content


Eclipse Community Forums
Forum Search:

Search      Help    Register    Login    Home
Home » Archived » B3 » Re: Multiphase creation of update site?
Re: Multiphase creation of update site? [message #556384] Wed, 01 September 2010 06:23 Go to next message
Eclipse UserFriend
Originally posted by: dhendriksREMOVE_THIS.tueREMOVE.THIS.nl

Hello all,

After being redirected to b3 from Buckminster, my question is whether
the scenario I presented (see below) is supported by b3. That is, does
(or will) b3 support automation of multi-phase build processes?

If so, what is the status of the implementation; is it ready for use in
real-life projects (as I read that b3 is still in the proposal phase:
http://wiki.eclipse.org/PDE/Incubator/b3/Proposal)

Thanks,
Dennis

Henrik Lindberg wrote:
> Hi,
> this is one of the goals of Eclipse b3 as it has a lot more
> compositional flexibility, and has unified all of the artifacts used by
> Buckminster into one.
>
> Don't think this will be supported directly in Buckminster even when
> used as an engine behind the scenes.
>
> Regards
> - henrik
>
> On 7/16/10 3:18 PM, Dennis Hendriks wrote:
>> Hello,
>>
>> For my QVTo project, I would like to create a p2 update site. The QVTo
>> file cannot be parsed if I don't have the metamodel it depends on
>> installed. Also, it requires a blackbox library, provided by an other
>> Java plug-in. Note that it is not enough to have the projects in the
>> workspace, I actually need the distributed jars installed into headless
>> Buckminster.
>>
>> What I do now, is create an update site for the metamodel/EMF project,
>> and the blackbox library Java plug-in project. Then install those
>> plug-ins into headless Buckminster. Then I create an update site for the
>> QVTo project, together with the other two projects.
>>
>> This requires multiple phases, and multiple intermediate/temporary
>> update sites. I needed to create a shell script and invoke Buckminster
>> with various commands, and different rmaps/cqueries for the multiple
>> phases. Is there any way to avoid this, that is, to automatically have
>> this resolved etc? I think it would be within the goal of Buckminster to
>> automatically do this for me, right?
>>
>> Dennis
>
Re: Multiphase creation of update site? [message #556514 is a reply to message #556384] Wed, 01 September 2010 15:33 Go to previous messageGo to next message
Henrik Lindberg is currently offline Henrik LindbergFriend
Messages: 2509
Registered: July 2009
Senior Member
The b3 project is found here: http://www.eclipse.org/modeling/emft/b3/

The original proposal was that b3 was to be a sub-project to PDE, but
since the ambitions goes beyond PDE, it was decided that modeling was a
better home. The PDE incubation part is still valid to contain
changes/support for b3 from a PDE perspective.

The part of b3 that is ready for use is the b3 aggregator - which helps
with part of the problem - that of performing operations on p2
repositories (aggregating, filtering, re-authoring, etc).

Don't know if the aggregator will help you all that much, as in your
problem there are building steps inbetween where the buckminster queries
and commands etc. are now executed.

What you want to achieve is not directly supported (as you found out)
and is typically performed with several phases orchestrated by a script
or using Hudson. I agree that it would be really nice to be able to do
all of the required steps "at once", and support declarative
specification of such a flow directly in b3 - i.e. the result of one
build operation is a "processor/compiler" in subsequent operations. Will
log an enhancement request for that for b3.

Regards
- henrik

On 9/1/10 8:23 AM, Dennis Hendriks wrote:
>>>
>>> For my QVTo project, I would like to create a p2 update site. The QVTo
>>> file cannot be parsed if I don't have the metamodel it depends on
>>> installed. Also, it requires a blackbox library, provided by an other
>>> Java plug-in. Note that it is not enough to have the projects in the
>>> workspace, I actually need the distributed jars installed into headless
>>> Buckminster.
>>>
>>> What I do now, is create an update site for the metamodel/EMF project,
>>> and the blackbox library Java plug-in project. Then install those
>>> plug-ins into headless Buckminster. Then I create an update site for the
>>> QVTo project, together with the other two projects.
>>>
>>> This requires multiple phases, and multiple intermediate/temporary
>>> update sites. I needed to create a shell script and invoke Buckminster
>>> with various commands, and different rmaps/cqueries for the multiple
>>> phases. Is there any way to avoid this, that is, to automatically have
>>> this resolved etc? I think it would be within the goal of Buckminster to
>>> automatically do this for me, right?
>>>
>>> Dennis
>>
Re: Multiphase creation of update site? [message #556634 is a reply to message #556514] Thu, 02 September 2010 07:38 Go to previous message
Eclipse UserFriend
Originally posted by: dhendriksREMOVE_THIS.tueREMOVE.THIS.nl

Hello Henrik,

I guess I'll stick to Buckminster for now and do a manual two phase
build from a Linux shell script.

Thanks for the detailed explanation!

Regards,
Dennis

Henrik Lindberg wrote:
> The b3 project is found here: http://www.eclipse.org/modeling/emft/b3/
>
> The original proposal was that b3 was to be a sub-project to PDE, but
> since the ambitions goes beyond PDE, it was decided that modeling was a
> better home. The PDE incubation part is still valid to contain
> changes/support for b3 from a PDE perspective.
>
> The part of b3 that is ready for use is the b3 aggregator - which helps
> with part of the problem - that of performing operations on p2
> repositories (aggregating, filtering, re-authoring, etc).
>
> Don't know if the aggregator will help you all that much, as in your
> problem there are building steps inbetween where the buckminster queries
> and commands etc. are now executed.
>
> What you want to achieve is not directly supported (as you found out)
> and is typically performed with several phases orchestrated by a script
> or using Hudson. I agree that it would be really nice to be able to do
> all of the required steps "at once", and support declarative
> specification of such a flow directly in b3 - i.e. the result of one
> build operation is a "processor/compiler" in subsequent operations. Will
> log an enhancement request for that for b3.
>
> Regards
> - henrik
>
> On 9/1/10 8:23 AM, Dennis Hendriks wrote:
>>>>
>>>> For my QVTo project, I would like to create a p2 update site. The QVTo
>>>> file cannot be parsed if I don't have the metamodel it depends on
>>>> installed. Also, it requires a blackbox library, provided by an other
>>>> Java plug-in. Note that it is not enough to have the projects in the
>>>> workspace, I actually need the distributed jars installed into headless
>>>> Buckminster.
>>>>
>>>> What I do now, is create an update site for the metamodel/EMF project,
>>>> and the blackbox library Java plug-in project. Then install those
>>>> plug-ins into headless Buckminster. Then I create an update site for
>>>> the
>>>> QVTo project, together with the other two projects.
>>>>
>>>> This requires multiple phases, and multiple intermediate/temporary
>>>> update sites. I needed to create a shell script and invoke Buckminster
>>>> with various commands, and different rmaps/cqueries for the multiple
>>>> phases. Is there any way to avoid this, that is, to automatically have
>>>> this resolved etc? I think it would be within the goal of
>>>> Buckminster to
>>>> automatically do this for me, right?
>>>>
>>>> Dennis
>>>
>
Re: Multiphase creation of update site? [message #611623 is a reply to message #556384] Wed, 01 September 2010 15:33 Go to previous message
Henrik Lindberg is currently offline Henrik LindbergFriend
Messages: 2509
Registered: July 2009
Senior Member
The b3 project is found here: http://www.eclipse.org/modeling/emft/b3/

The original proposal was that b3 was to be a sub-project to PDE, but
since the ambitions goes beyond PDE, it was decided that modeling was a
better home. The PDE incubation part is still valid to contain
changes/support for b3 from a PDE perspective.

The part of b3 that is ready for use is the b3 aggregator - which helps
with part of the problem - that of performing operations on p2
repositories (aggregating, filtering, re-authoring, etc).

Don't know if the aggregator will help you all that much, as in your
problem there are building steps inbetween where the buckminster queries
and commands etc. are now executed.

What you want to achieve is not directly supported (as you found out)
and is typically performed with several phases orchestrated by a script
or using Hudson. I agree that it would be really nice to be able to do
all of the required steps "at once", and support declarative
specification of such a flow directly in b3 - i.e. the result of one
build operation is a "processor/compiler" in subsequent operations. Will
log an enhancement request for that for b3.

Regards
- henrik

On 9/1/10 8:23 AM, Dennis Hendriks wrote:
>>>
>>> For my QVTo project, I would like to create a p2 update site. The QVTo
>>> file cannot be parsed if I don't have the metamodel it depends on
>>> installed. Also, it requires a blackbox library, provided by an other
>>> Java plug-in. Note that it is not enough to have the projects in the
>>> workspace, I actually need the distributed jars installed into headless
>>> Buckminster.
>>>
>>> What I do now, is create an update site for the metamodel/EMF project,
>>> and the blackbox library Java plug-in project. Then install those
>>> plug-ins into headless Buckminster. Then I create an update site for the
>>> QVTo project, together with the other two projects.
>>>
>>> This requires multiple phases, and multiple intermediate/temporary
>>> update sites. I needed to create a shell script and invoke Buckminster
>>> with various commands, and different rmaps/cqueries for the multiple
>>> phases. Is there any way to avoid this, that is, to automatically have
>>> this resolved etc? I think it would be within the goal of Buckminster to
>>> automatically do this for me, right?
>>>
>>> Dennis
>>
Re: Multiphase creation of update site? [message #611624 is a reply to message #556514] Thu, 02 September 2010 07:38 Go to previous message
Eclipse UserFriend
Originally posted by: dhendriksREMOVE_THIS.tueREMOVE.THIS.nl

Hello Henrik,

I guess I'll stick to Buckminster for now and do a manual two phase
build from a Linux shell script.

Thanks for the detailed explanation!

Regards,
Dennis

Henrik Lindberg wrote:
> The b3 project is found here: http://www.eclipse.org/modeling/emft/b3/
>
> The original proposal was that b3 was to be a sub-project to PDE, but
> since the ambitions goes beyond PDE, it was decided that modeling was a
> better home. The PDE incubation part is still valid to contain
> changes/support for b3 from a PDE perspective.
>
> The part of b3 that is ready for use is the b3 aggregator - which helps
> with part of the problem - that of performing operations on p2
> repositories (aggregating, filtering, re-authoring, etc).
>
> Don't know if the aggregator will help you all that much, as in your
> problem there are building steps inbetween where the buckminster queries
> and commands etc. are now executed.
>
> What you want to achieve is not directly supported (as you found out)
> and is typically performed with several phases orchestrated by a script
> or using Hudson. I agree that it would be really nice to be able to do
> all of the required steps "at once", and support declarative
> specification of such a flow directly in b3 - i.e. the result of one
> build operation is a "processor/compiler" in subsequent operations. Will
> log an enhancement request for that for b3.
>
> Regards
> - henrik
>
> On 9/1/10 8:23 AM, Dennis Hendriks wrote:
>>>>
>>>> For my QVTo project, I would like to create a p2 update site. The QVTo
>>>> file cannot be parsed if I don't have the metamodel it depends on
>>>> installed. Also, it requires a blackbox library, provided by an other
>>>> Java plug-in. Note that it is not enough to have the projects in the
>>>> workspace, I actually need the distributed jars installed into headless
>>>> Buckminster.
>>>>
>>>> What I do now, is create an update site for the metamodel/EMF project,
>>>> and the blackbox library Java plug-in project. Then install those
>>>> plug-ins into headless Buckminster. Then I create an update site for
>>>> the
>>>> QVTo project, together with the other two projects.
>>>>
>>>> This requires multiple phases, and multiple intermediate/temporary
>>>> update sites. I needed to create a shell script and invoke Buckminster
>>>> with various commands, and different rmaps/cqueries for the multiple
>>>> phases. Is there any way to avoid this, that is, to automatically have
>>>> this resolved etc? I think it would be within the goal of
>>>> Buckminster to
>>>> automatically do this for me, right?
>>>>
>>>> Dennis
>>>
>
Previous Topic:Re: Multiphase creation of update site?
Next Topic:Hudson Support
Goto Forum:
  


Current Time: Tue Mar 19 10:21:45 GMT 2024

Powered by FUDForum. Page generated in 0.23313 seconds
.:: Contact :: Home ::.

Powered by: FUDforum 3.0.2.
Copyright ©2001-2010 FUDforum Bulletin Board Software

Back to the top