Forum OpenACS Development: Portal port from dotLRN to OpenACS

Posted by Michael Hinds on 03/20/03 12:23 PM

I'm in the process of trying to work out how much work needs to be done to get the new-portal package from dotLRN into OpenACS. I've discovered a dependency on static-portlet, which itself depends on dotlrn. After a quick look at static-portlet, I found that it uses procs from dotlrn_community and dotlrn_applet, but on the face of it this doesn't seem to be a heavy dependency. So it looks like I should try to port static-portlet, hacking out the dotLRN bits, as part of the whole portal porting effort. Or should I make the portal independant of static-portal?

2: Re: Portal port from dotLRN to OpenACS (response to 1)

Posted by Peter Marklund on 03/20/03 01:04 PM

Michael,
have you looked at all at the portal package (the refactored new-portal package by OpenForce) in the OpenACS repository and how it differs from new-portal?

The important thing to me would be that we focus on either of these two packages (throw out the other one) and make this package work in both OpenACS and dotLRN.

Great news that you've taken this work!

3: Re: Portal port from dotLRN to OpenACS (response to 1)

Posted by Michael Hinds on 03/20/03 04:48 PM

Peter,

Yes, those are the two I'm comparing. Some changes are obviously updates to the latest way of doing things (namespaces, etc.) but in some cases it's not immediately clear what the reason for the change was. I don't know whether the OpenForce guys were in the middle of moving stuff around, what decisions they've already made, whether they were implementing new features at the same time...

I should add that while I'm eager to get cracking on the port myself, at this stage I'm only really looking at how long it might take. I can't afford to be sucked into a black hole of work. I reckon if it looks like being more than 2 weeks worth I might have to consider other options. Damn these commercial realities!

4: Re: Portal port from dotLRN to OpenACS (response to 1)

Posted by Don Baccus on 03/20/03 06:19 PM

We definitely want to use the refactored version, i.e. to finish it. I'd suggest ignoring new-portal from dotLRN at the moment and concentrate on getting the refactored version running standalone.

As I mentioned earlier in e-mail, I was planning to pick this up myself in a few weeks, so even if you don't get finished in the time you have available the work you do will be a big help.

We definitely don't want a dependency on static portlet but I'm willing to be the person to figure out how to re-integrate the refactored portals package to dotLRN.

Or not, to be honest, if it looks like an incredible PITA to do so (the upgrade scripts being the biggest PITA)

While I don't want us to have both versions if at all possible, the most important thing to me is to get the refactored version working properly as a package integrated into OpenACS 4 without dotLRN. The refactored version is a lot cleaner and the datamodel's somewhat lighter (they only made things objects which really need to be objects, which can save a considerable amount of space on an installation like Sloan which has thousands of users each with their own personal portal)

5: Re: Portal port from dotLRN to OpenACS (response to 4)

Posted by Denis Roy on 03/20/03 08:45 PM

Don,

we are also interested in having a stand-alone portal package. In fact, if we can somehow fit it into our client's timeline, we might end up doing it rather sooner than later.

Can you give me a short overview of what needs to be done and a conservative estimation on how long it would take to finish the refactored portal package?

6: Re: Portal port from dotLRN to OpenACS (response to 1)

Posted by Don Baccus on 03/20/03 08:59 PM

The first task Michael has agreed to take on is to do exactly that: to try to inventory what needs doing and to come up with a rough estimate of the time required.

I'd like to thank Michael for doing this in public, BTW, it's been lying there ever since Open Force declared they didn't have the time or resources to finish it up last fall. It's great to see someone step up to take some preliminary steps towards completion. And thanks for the offer of Sussedorf & Roy help.

After Michael comes back with more info we can put our heads together and decide how best to finish the work.

How does that sound?

7: Re: Portal port from dotLRN to OpenACS (response to 6)

Posted by Denis Roy on 03/20/03 09:10 PM

Sounds great. I really hope Michael will come back with some useful information quite soon.

I would love to offer some help here but our general problem is that we don't have too much time left before we have to take a decision on whether we want to go for the refactored portal or not. To be honest, the idea to use the refactored portal instead of dotLRN just came up today although we are already expected to come up with some concrete timelines. Let's see how we can make it fit.

8: Re: Portal port from dotLRN to OpenACS (response to 1)

Posted by Don Baccus on 03/20/03 09:49 PM

The expectation is that we'll provide upgrade scripts unless it's just too damned hard :(

But I don't think it will be, the table structures are the same, roughly, just need to "de-objectify" them to upgrade and I've already give though to it.