On Mar 21, 2008, at 5:35 PM, Sergiu Dumitriu wrote:
Vincent Massol wrote:
Hi Asiri,
I'm personally not in favor of having those 2 ideas done as GSOC projects. My reasons: * They are the most complex possible since they are at the very heart of XWiki * They need to be done by existing committers since this is at the heart of xwiki and we'll need to support them on the long run * They are NOT standalone projects and they interact too much with other parts of XWiki. * They have not had the experience of implementing the first generation version and thus will not know the pitfalls and all the ideas that have been raised on these topics over the years
In essence this means that almost any line written will need to be discussed at length, validated, understood by existing committers. I don't think it's fair to give this to a GSOC student especially as it'll be almost impossible for him/her to be sure of delivering anything. Then it would mean several mentors and that they are 100% available which is not going to happen. And last if they were I would prefer that they tackle it themselves since again this is really at XWiki's heart.
This is my opinion only and Segiu obviously has a different one since he proposed these 2 projects. So first thing, we need to decide if we want to have them a GSOC project.
They are definitely hard, so they aren't for anybody. But I thought that we could use an extra hand, probably not a BSc student (since GSoC accepts MSc and PhD students, too), someone with a lot of experience, with whom we would work together, instead of just supervising. For "younger" students, we have lots of other fun projects.
That would only work if the mentor was working on the same topic at the same time IMO. And if that's the case it might be more interesting that the mentor does it by himself with the community. It's not like these are work-intensive operations that require lots of people. They are really design-intensive and require lots of careful design decisions. I know that if I personally won't be working on this at the same time as the student is then it's unlikely I'll have the time to review the design since it really requires to be focused and "inside" the topic to be able to comment/review. And since these are ultra critical topics I think we will need everyone's input and since it may take time to come I don't feel it's doing a service to the student since he won't be able to progress quickly and may not even be able to finish his SOC. -Vincent
What do other committers think?
Thanks -Vincent
On Mar 12, 2008, at 4:12 PM, Asiri Rathnayake wrote:
Hi Vincent & All,
This is about the above two GSoC ideas,
I would like to know what are the deliverables for the above two projects ? My guess is,
1. Finished Components (i think these are going to be two jar files ?, i'm yet to discover plexus)
2. Test cases.
3. Current core integrated with the two components.
4. Anything else ?
Since the current core doesn't follow a components based architecture, what kind of complications will arise in the integration phase ?
Vincent, what about the two implementations of the data-model component in the sandbox ? have you made a decision ? :)
Your help and comments are very much appreciated.