[Collozeum] How to release via Mogul?

[Filip Konvička]

Hi again,

> Dear all,
>
> released Collozeum projects should be easily accessible via Mogul. I 
> have no experience with Mogul releases yet. The process is described 
> at http://www.mozart-oz.org/mogul/node8.html#chap.contrib, but I 
> recall earlier posts here which described certain problems like that 
> it was meanwhile hard to get a Mogul ID (couldn't find the email 
> anymore..). Could someone just summarise recent experiences? Filip?

from my point of view the current process is rather complicated. The 
problem is that the MOGUL librarian actively downloads the libraries 
from some host sites. So this is (as Denys said in the past) the 
pull-model, and requires that the contributors maintain public web sites 
with the libraries (I use a free public web host) and also there must be 
an admin that assigns mogul IDs (currently probably Denys again).

I think that the better way would be hosting the libraries at a common 
place, like gforge, and re-writing the librarian to support the push-model.

> More specifically, what should be the mogul root id for Collozeum 
> projects? Should every Collozeum contributor become registered at 
> MOGUL and get a MOGUL ID? Or would it be sufficient (for now at least) 
> if only Collozeum is registered as a meta author, and further authors 
> are then just listed with their names as strings instead of MOGUL IDs?

IMHO, it would suffice if the project got a meta id. You would just need 
to ensure that the contributors adhere to some common license.

If you want, I can show you the build makefiles that I use for creating 
the distribution packages. Once created, the distribution packages must 
be uploaded to the web host that the MOGUL librarian knows of and from 
which it will download the new versions during the nightly runs.

>
> Where should release files be stored? I assume it would be best to use 
> the Gforge release mechanism for storing file releases and just 
> telling ozmake/MOGUL where to find these.

You need to make the files accessible via http, so if you can do this 
within Gforge this would be probably the best option. My idea of the 
process is that each contributor maintains her package(s) and uploads 
just those packages to the repository, as needed, not touching the others.

Cheers,
Filip