[theforum] weo 2b: to OSS or not?

David Kaufman david at gigawatt.com
Mon Nov 8 13:52:16 CST 2004


Madhu Menon <webguru at vsnl.net> wrote:
> There's also the issue of dumping all the development knowledge
> accumulated over the years that we've spent working on the current
> system. During that time, we've probably encountered issues that
> seemed trivial but were a PITA later, etc. The "start from scratch"
> approach is not always a good one, especially if you're not going to
> involve the original developer(s). (see
> http://www.joelonsoftware.com/articles/fog0000000348.html )
> Apart from customising whatever software you pick, you will also have
> to migrate the existing data - articles, users, comments, ratings,
> etc. without losing any of it. Is everyone here confident it will all
> go without a hitch?

That was exactly my point, too.  If we expect to move off of the current 
server in 90 days, we don't really have time to dick around debating new 
features, new publishing workflows, and writing new software.

I'm normally the loudest evangelical proponent of OSS solutions, but in 
this case, for the project to succeed on time (i.e. by January 31st, 
which everyone just voted was the plan), I think we must consider the 
easiest, quickest option, which is to just run the same system we have 
now, by porting it to ColdFusion for Linux, which is not as big a job as 
you might think... Two years ago, I ported thee fairly huge high traffic 
sites, AsiaSource.org, AsiaBusinessToday.org and AsiaFood.org from 
Netscape Server and Oracle on Solaris, to Apache and MySQL on Linux. 
It's still ColdFusion to this day, but that now runs on an open source 
operating system, webserver and database.

It took about 2 months, full-time, and the day we relaunched (in Jan 
2002) the sites all looked and acted exactly the same as they did before 
the switchover, but Asia Society has saved tens of thousands in 
dedicated Sun machine hosting, Oracle and Netscape server licenses.

> And lastly, how are we already making software recommendations without
> putting down actual functional requirements? I'd really like to see
> that first.

+1 to that!

This is why I'm pushing for the no-spec spec.  There will be ~plenty~ of 
issues to resolve in this short period of time, just moving these sites, 
all the lists and their archives and settings, and getting the *current* 
content publishing backend working *without* changing the functionality, 
too.

I fear that much of the current debate *presumes* that a complete 
rewrite is GOING to be performed before 31 Jan, and seeks to pick a 
platform first, and then spec new features for a ground up rewrite.  To 
me, that's Back Asswards!  First, you define features and functionality 
needed (the goals).  Only then can you devise a strategy for meeting 
those objectives, like choosing a technology platform.  Then you remove 
all tasks which do not directly support the goal.

The goal is what is being muddied in these waters.  I think that the 
goal should be limited right now, to JUST moving leo, weo, deo and beo 
(did i miss any?) successfully to our ServerMatrix machine on schedule. 
Then we can start rebuilding!  To make the deadline I really do think we 
have to limit the scope to simply matching the functionality we already 
enjoy, and not than beginning a nebulous and ambitious redesign of the 
entire system from it's underlying language to it's workflow and 
"business rules".

C'mon guys, I like Drupal too, and this discussion represents great 
progress that is very important to the long-term success of evolt, but 
it is not urgent.  Moving, on time is urgent and important.  Can we 
really hope to iron out all the new features being proposed, write all 
this new code, integrate all these OSS tools, fix the inevitable bugs 
and problems and convert all of our existing data, all while also moving 
to a new O/S and hosting provider?  Isn't that like converting your car 
from gasoline to hydrogen, while driving it down the highway?

Let me repost Madhu's very relevant link that also sums up my (obviously 
strong) opinions on this matter: 
http://www.joelonsoftware.com/articles/fog0000000348.html

Joel hits the nail right into our heads, methinks :-)

-dave 



More information about the theforum mailing list