"Chanining" resolvers [message #669715] |
Wed, 11 May 2011 09:11  |
Eclipse User |
|
|
|
I want to include a feature and it's plugins in my product.
The feature and plugins are stored into a .jar file on artifactory.
Also, the .jar file doesn't contain a p2 update site, but a classic (3.4 style).
The structure of the .jar containing the resources is:
- .
---- features/
------- my.feature_1.0.0.20110511-1514.jar
---- plugins/
------- my.plugin_1.0.0.20110511-1514.jar
---- MANIFEST.MF
My questions are:
First of all, is it possible to use Buckminster to import from this source even by developing some sort of resolver or extending a reader type?
Is it possible to chain the resolvers somehow. For example:
Use a p2SiteGenerator reader which uses a zip reader which uses an artifactory reader?
What would be the best practice for doing such chaining (importing stuff into the workspace as well as in the build)?
|
|
|
Re: "Chanining" resolvers [message #669725 is a reply to message #669715] |
Wed, 11 May 2011 09:27   |
Eclipse User |
|
|
|
On 2011-05-11 11:11, Fabian Baboschi wrote:
> I want to include a feature and it's plugins in my product.
>
> The feature and plugins are stored into a .jar file on
> artifactory.
> Also, the .jar file doesn't contain a p2 update site, but a
> classic (3.4 style).
> The structure of the .jar containing the resources is:
> - .
> ---- features/
> ------- my.feature_1.0.0.20110511-1514.jar
> ---- plugins/
> ------- my.plugin_1.0.0.20110511-1514.jar
> ---- MANIFEST.MF
>
> My questions are:
> First of all, is it possible to use Buckminster to import
> from this source even by developing some sort of resolver
> or extending a reader type?
>
Yes, you can probably do that out of the box using a url reader type with a matcher (to extract names and versions from
file names).
> Is it possible to chain the resolvers somehow. For example:
>
> Use a p2SiteGenerator reader which uses a zip reader which
> uses an artifactory reader?
>
Yes, either put them both in the same provider (best fit wins) or put them in separate providers and use locators with
failOnError=false (first fit wins).
> What would be the best practice for doing such chaining
> (importing stuff into the workspace as well as in the
> build)?
>
I would never import binaries into the workspace. Put them in he target platform.
Regards,
Thomas Hallgren
|
|
|
Re: "Chanining" resolvers [message #669793 is a reply to message #669725] |
Wed, 11 May 2011 14:23   |
Eclipse User |
|
|
|
Actually, after looking closer at the content of the .jar file, the exact structure is:
---- features/
------- my.feature_1.0.0.20110511-1514/
---------- feature.xml
---- plugins/
------- my.feature.plugin1_1.0.0.20110511-1514/ (and subfolders)
------- my.feature.plugin2_1.0.0.20110511-1514.jar
---- META-INF/MANIFEST.MF
So it's an "ancient" feature (that is used by another 3.4 build).
Can the p2 reader be used on the folder structure above?
Can I import that into the target platform?
Also, by chaining, what I meant is for the output of the first reader to be used as input for the second reader and so on.
In my case, my problem is exactly this:
The maven2 reader should output the jar file containing the above structure, then a url.zipped reader should use the previous output and have another output that would then be used by a third reader (eclipse.import?, p2?) that is able to parse the structure I showed before.
|
|
|
Re: "Chanining" resolvers [message #671240 is a reply to message #669793] |
Tue, 17 May 2011 09:08  |
Eclipse User |
|
|
|
Can I execute an action before the resolution of the dependencies for a feature is performed?
In my scenario, I'd like to unpack the jar file containing the folders above when a referring feature needs it, but before the actual resolution is performed. Maybe this way the resolution will find it when it searches for it (via a rmap local entry pointing to the location where this action will output the result).
I tried using generators, but it just doesn't work. There is no error in the logs, but I get the dreaded "A selected specification is unresolved".
|
|
|
Powered by
FUDForum. Page generated in 0.06923 seconds