Forum OpenACS Q&A: Re: .LRN Portal Framework in OACS?

Collapse
Posted by Don Baccus on
Well, not only can one package offer multiple portlets, but any package can offer a portlet to render the contents of another package (like forums) that offers a comprehensive datamodel.

See the recent thread in the OpenACS development forum.

As far as integrating portals into OpenACS proper ... originally Open Force started a rewrite/improvement before they dropped out of the community over our power struggle/war and later Ben dissolved the company, before the rewrite was finished.

I looked last summer into completing it and got it nearly to the point where page rendering worked then got distracted with other (paying) work.

At this point, in our OCT meeting this morning (total coincidence) I suggested we just improve the .LRN new-portal package and abandon the rewrite.  The existing code, though not beautiful, appears to work well and scale reasonably, and with so much existing code depending on it, switching to the rewritten version would require complex and maybe dicey upgrade scripts.

So we're going to live with the first cut .LRN portal package.

Perhaps a good argument for good design and implementation the first time?  The "plan to throw the first one away" argument is a good one but only for those who don't actually have users or customers using the first cut code.  A lesson, perhaps, that if you don't want to be stuck with your prototype forever, don't release it but do your improved version and release that.

Whatever ... the existing new-portal package is very functional and though not beautiful we can deal with that.  Including incremental improvements over time that don't flat-out break code that relies on it.

So. Bob, is it true you're going to buy me an EOS 1DS for xmas?  No?  Damn!

Collapse
Posted by Dirk Gomez on
Also read the Mythical Man Month - recently reissued me thinks. It has a good chapter on what happens if you throw away the first design because it was too crappy. It is chapter 5 - The second system effect (to which I and others had refered to a couple of times)

The .LRN portal framework code is awful and contracted, but it cuts it for the user. We should stick to it for the time being until someone has a compelling replacement.

As it is very unlikely that someone will have a compelling replacement we should continue to work on the current code :)