[infinispan-dev] Building the Infinispan Server

[Radim Vansa]

I agree with Sanne that the build should work out of the box - git clone 
&& mvn install must not fail.
As now the cache-store lifecycle is independent, infinispan-server in 
snapshot version should not consume snapshot version of cache stores but 
a released one. If one wants to use snapshot versions of cachestores, he 
should build it with profile which would switch that.

Radim

On 11/06/2013 09:09 PM, Sanne Grinovero wrote:
> On 6 November 2013 18:58, Tristan Tarrant <ttarrant at redhat.com> wrote:
>> On 11/06/2013 03:55 PM, Sanne Grinovero wrote:
>>> Right I misread the error, sorry, but still it's a broken flow.
>>>
>>> Would be nice to have a build tool which knows how to build the whole
>>> project without me going to hunt down 3 different repositories.
>>> It was absurd enough to need to apply two pull requests on two
>>> different repositories to resolve one JIRA (or actually to handle a
>>> single pull request).
>>> #fail
>> The real fail is that you didn't send a PR with such a build tool :)
> No offence, but if it doesn't build out of the box, I'm not going to touch it.
>
> The contributor barrier is high enough that it doesn't need more overhead,
> people not working fulltime on Infinispan can't handle this.
>
> I'm not asking you to make a new tool but pointing out it was a
> mistake to split out the project in many repositories, if there is an
> assumption that this high level of coupling is acceptable.
> Rather than splitting CacheStores out I'd have merged the server project.
>
> It's certainly doable to have projects managed in different
> repositories and having different lifecycles (like our other projects
> do integrate with Infinispan), but there usually is no expectation of
> keeping them aligned: we develop SPIs or APIs and make plans across
> projects to handle this complexity: I had warned about this kind of
> trouble when the split was suggested and am not particularly happy to
> see that my most pessimistic views became a reality.
>
> We can have separate repositories, but it needs to be managed
> differently.
> We'll have to make sure different pull requests are sent in the
> correct order to each project, that issues are scheduled accordingly
> on the independent release cycle of each component, and that tags are
> released to Nexus to allow for developers to focus on one module at a
> time: having "misaligned snapshots" is another source of confusion
> that I could have had in this scenario.
>
> Sanne
> _______________________________________________
> infinispan-dev mailing list
> infinispan-dev at lists.jboss.org
> https://lists.jboss.org/mailman/listinfo/infinispan-dev


-- 
Radim Vansa <rvansa at redhat.com>
JBoss DataGrid QA