Forum OpenACS Q&A: Trying to catch up....Help?

Hi,

I started looking at ACS a couple years ago...but never did much with
it after I downloaded it. I now have the time to rectify this mistake
and experiment a bit.  Problem is, I'm having trouble keeping up with
what is going on now.

I kept up with the aD fiasco (morbid curiosity), but I don't
understand much about the branching of the toolkit (OpenACS,
OpenNSD...?), what modules exist, what modules work with what
toolkits, what setup would be best for my needs, etc.???

I'm only interested in Tcl branches, as I've had more than my fill of
Java...

I realize much of this information is scattered around the forums,
but I would appreciate it if someone gave me an overview.  I'm rather
dim-witted technically and I don't want to waste hundreds of hours
trying to get incompatible things to work together simply because I
made idiotic newbie assumptions.

I'm primarily interested in the bulletin board, e-commerce, and CMS
capabilities.

I'm as close to a total newbie as one can be, so suggestions that are
robust (less hacking and tweaking) and offer gentler learning curves
are most welcome.

Many thanks in advance!

- David

OTHER INFO:
I have a rack of 6 Apple xserves, so I'm especially interested in
things on the Mac OS X front, but I can buy a computer and re-learn
FreeBSD if absolutely necessary? [Suggestions?] I also have no
experience with PostgreSQL, but own a Frontbase db license (which has
a Tcl module.) Should I spend my time learning Postgres or trying to
get the data model to work with Frontbase?  Also, I have quite a bit
of time/money invested in the Zeus webserver (Yes, I prefer open-
source, but it has great speed and virtual server support!), does
anyone have any experience running Zeus alongside AolServer?

You can also run OpenACS on Mac OS X, although it's not really
ready for production yet. I do my development work on a
Powerbook, and it works just fine. There is a thread somewhere
here about getting set up -- if you're a newbie, I suggest just
installing the prepackaged deal (I don't remember who did it, but
it sure was a lot easier). The documentation for Mac OS X
doesn't work for Mac OS 10.1, unfortunately. Of course, if you
have some time on your hands, we'd all love it if someone would
get the documentation working...

If you are going to buy a computer, you should probably install
Linux on it. FreeBSD would work, but there are a lot of Linux
people here to help out :)

Hi David,

Furfly has been doing work on ecommerce and apparently has
some code they will be releasing soon: You can find out the
latest here:

https://openacs.org/bboard/q-and-a-fetch-msg.tcl?msg_id=0004o4&topic_id=12&topic=OpenACS%204%2e0%20Design

OpenForce is working hard on a major upgrade to bboard, as
you'll see here:

https://openacs.org/bboard/q-and-a-fetch-msg.tcl?msg_id=0004ir&topic_id=OpenACS%204%2e0%20Design&topic=12

CMS capabilities are a little murkier. Although aD's underlying
content repository architecture is alive and well, the community
basically decided to throw out the CMS itself because it had
serious problems and nobody was particularly enthusiastic
about wading into the code to fix it. A new, much simpler CMS
module called "Edit This Page" (or ETP) was developed by
Musea. They are actively working on evolving it, as you'll see
here:

https://openacs.org/bboard/q-and-a-fetch-msg.tcl?msg_id=00047q&topic_id=12&topic=OpenACS%204%2e0%20Design

Here's a thread on using OpenACS on OS X:

https://openacs.org/bboard/q-and-a-fetch-msg.tcl?msg_id=0002mP&topic_id=11&topic=OpenACS

Regarding the question of buying a new box, it's certainly true
that you'll get the quickest and easiest answers from the
community if you're running Linux just because so many people
are doing it. However, there is a fair amount of interest in OS X. I
personally would love to hear performance reports on XServe.

Thanks for the clarification, Don.

To be fair, I don't think Musea was setting out to create a full-on
CMS replacement. They had simpler needs and developed a
system that met those needs. ETP2 looks like it aims to be more
sophisticated but still falls well short of the ambitions that CMS
attempted to fulfill. The fact is that if you're looking for a
full-featured, enterprise-appropriate CMS, then OpenACS doesn't
have one that's in good working order and nobody that I know of
has publicly announced that they're working on one.

Thanks for the incredibly helpful posts...keep 'em coming!

Jade:  I'm also looking to do development on a TiBook.  I usually prefer *BSD to Linux, but I guess I'll go with either Linux or OS X.
I think I read somewhere on this site that Red Hat is the preferred distro of Linux (if I end up chickening out on the OS X)?

Russell:  I'd love to be updated about your progress with an OpenACS/OS X combo in a production environment as you work with your client.  I'll try to reciprocate, although I'm sure I'll make MUCH slower headway...my company runs almost everything on OS X, so I really want to be able to get something working.

Michael:  Thank you! You're one heckuva tour guide.  I would love to see OpenACS scream on an Xserve (everything else I've tried on Xserve does, in fact, go Mach X ;) )

Didn't you mention pgMac on another post?
Have you (or anyone else) actually tried it? - I seem to be their ideal customer (I'm stupid/inexperienced and have $250.)

I would think a decent CMS would be a higher priority than it is, since it is such a common need?

Andrew:  Last time I looked at Postgres it was not anywhere near Frontbase in functionality/ease of use (Of course, I know they've both made quantum leaps since then...)

I won't actually make an argument that Frontbase is better than anything, since I'm not experienced enough to fully understand the subtle technical merits of RDBMS.  I will, however, ask you a question:  If you could use either Postgress OR Oracle - free of all charges, which would you choose?  Oracle, right?

At my last start-up I hired a senior Oracle DBA, an OCP who had worked for Oracle.  He ended up choosing Frontbase over Oracle 8, even though we were budgeted for Oracle and had already purchased Oracle training contracts for our junior DBAs.(!)  This was after extensive testing, BTW.

We saved enough on db licenses, training, and overtime to...well, we saved MUCHO.  (Of course, we were implementing in WebObjects - and a few servlets - so the Frontbase EOF support saved us a great deal of time additionally.)

Of course, this is just an anecdotal thing - but I thought I would relay our experiences so that you wouldn't dismiss Frontbase out of hand if you ever have to work on OS X...their website is at frontbase.com.

We have no legacy code in Frontbase for the apps I'm planning, but it would be trivial to have one of my employees or a third-party (i.e. a real programmer) write a db driver.  [I'm just a dabbler, but I have access to actual talent;) ]

I guess since I don't have a good reason to use Frontbase, I'll learn Postgres.  Any Postgres book recommendations? The Momjian book gave me a headache once ;)

Now that I've mentioned WebObjects, I just have to ask:  would there be any (likely to happen) way to implement something like EOF and/or the session management methods of WO in OpenACS?  (Cookies are UGLY!)

Please be kind, remember that I'm ignorant;)

David, I've never used pgMac and I don't know anybody who has,
though it's possible I posted a link to it if I saw it somewhere at
sometime. (I honestly don't remember whether I did or not.)

Regarding the CMS, I think the problem is that it's a pretty large
undertaking--big enough that the typical (relatively small)
OpenACS shops probably couldn't just go out and put in the time
to do it right without fairly significant client sponsorship. Now, if
somebody had the dollars to fund the initial development, I think
you'd see a lot of community energy going into maintaining and
enhancing the output of that first sponsored effort.

That's just my own guess, though.

In regards to CMS.  A lot of the pieces are in place (the content-
repository data model, templating, workflow, and now new-portal
for aggregation of content), but they are still evolving.  The aD
CMS fit the cart before the horse cliche.  It uses several of these
tools and then provides a overarching admin interface for them.
Everyone seems to hate the CMS for its interface, but the
overlooked truth is the CMS was responsible for getting a lot of
the needed tools off the ground.  The underlying tools are where
the value is and that is where you are seeing the most developer
interest.
I think we will reach a critical mass when the supporting
packages hit their stride.  Then it will make sense to tie them all
together.  Of course, it would be a mistake to develop them
without the goal of unified CM functionality in mind.

OpenACS 4.5's functionality goes beyond CM, but CM
functionality is present right now in the core of OpenACS.  I
wouldn't be surprised if we arrive at  a pre-configured vertical
app built on OpenACS called "dotCMS" or something.

If you're interested in making OpenACS more newbie-friendly to
install and administer, then you probably want to look at this
thread:

https://openacs.org/bboard/q-and-a-fetch-msg.tcl?msg_id=0004jo&topic_id=OpenACS%204%2e0%20Design&topic=12

<blockquote><i>Command line may be more efficient, but pretty pictures make Joe Schmo comfortable AND productive.</i></blockquote>
Mac OS X isn't the only Unix-based system that makes it unnecessary to use the CLI for most things.  Modern Linux distros have vastly improved in this regard in the last couple of years, along with autodetecting and supporting hardware.
<p>And under Mac OS X you'll still need to do some CLI work to get an OpenACS system up.  But I wouldn't worry about it, we're talking servers here and larger school districts are already running DNS, mail, web and various servers.  They're professional IT types - mine, the Multnomah County ESD most runs Linux these days anyway.
<p>Of course the situation changes at the level of the individual school or for small, rural districts.
<p><b>cookies vs. URL</b></p>
<p>Encoding state information in URLs rather than cookies makes search engine links and the like interesting.  Having an expired session state value show up in a search engine link isn't very cool, for instance.
<p>OpenACS tries to place a session state cookie on your machine but runs fine if you refuse it.  You only *need* cookies enabled to log in.  Cookies can be set to time out, which is useful in scenarios where you might log in from PC in, say, a library and forget to log out.  If you do remember to log out, the system clears the cookie.
<p>If your authentication information is encoded in URLs, then the library PC's browser history will allow anyone to come in and pose as you, even days later.  You could implement such a system so logging out made the old URLs not work (and time out older ones, too) but then you can't bookmark into the site ...
<p>The short story: cookies were invented for a reason, that reason being that trying to track state in URLs is inadequate.

urgh. piqued not peaked.  doh!

Matthew,

It is true that there is not cookbook of techniques to solve common tasks using the toolkit. But as you said the developers involved are knowledgable and helpful. If you have a question on how to approach a situation, post to the bboard. Chances are someone else has already had the same problem. If they have solved it, they can provide a solution, if not, you have one more person to help think about the solution.

So, don't be afraid to ask questions.

Furthermore, Matthew,

Dialog from answers to questions become notes for improving documentation --helping to shape documentation.  Questions are critical indicators to knowing where documentation is lacking.

Ask away...