Re: [buckminster-dev] Making Buckminster understand SVN branching

[Thomas Hallgren <thomas@xxxxxxx>]

Mikhail Kadan wrote:
> Hi.
>
>     Not sure what you mean by automatically here. Buckminster cannot
>     guess what branches you have in mind and in what order they should
>     be searched. You must specify this somewhere. You don't need to do
>     it once for each component of course. You can create an advisory
>     node that matches all components using the pattern '.*'.
>
>
> By "automatically" I mean that it may be some naming convention for 
> branches to "mark" them for Buckminster to search.
>
> For example, let's assume I have three branches for plA:
>
> R1.0.x
> R1.1.x
> testNewAPI
>
> If there is a naming convention saying "all branches that apply to 
> pattern RX.Y.x, where X and Y are numbers, must be included 
> automatically into search path", so the branches "R1.0.x" and "R1.1.x" 
> (as they follow this naming convention) will be included in search 
> automatically, and branch "testNewAPI" must be included manually.
Aha, yes. Buckminster can do branch/tag name to version conversion. I.e. 
it can extract the component version from a branch or tag name rather 
then from the meta-data of the component. You can create a provider that 
uses a <versionConverter> element. It uses two regexp substitutions, one 
to translate the branch to a version and one to do the opposite. 
Buckminster will then make an attempt to find the highest version. When 
this scheme is used, you don't use the branch/tag path and you don't use 
the trunk.

You can find more information here: 
http://wiki.eclipse.org/Buckminster_component_meta-data_language_%28Reference%29_-_LATEST#VersionConverter_component

Regards,
Thomas Hallgren