Forum OpenACS Q&A: Advice to ACS newbie wanting to do the problem sets, learn ACS

OS: RedHat 7.2 (because it's as good as anything else and most popular which means that you have the greatest chance to find the software/help from other people on the net).

If you don't want your box to be hacked, don't plug it into Internet or run behind a tight firewall e.g. in a hardware router with NAT. Really. If you don't have the expertise to secure your box then don't take risks or acquire the expertise.

If you just want to learn *nix, databases, programming, XML then expect to be a beginner after 5 years of hard work (http://www.norvig.com/21-days.html). In other words: don't expect anyone to give you a reasonable advice on achieving such broad goals because there is none (except maybe: work hard, read books, program a lot etc.). If you want reasonable advice set reasonable goals.

Whether you'll use PostgreSQL or Oracle, RedHat 7.2 or FreeBSD, ACS or OpenACS doesn't really matter. Just pick one and run. Fundamentals are important and they are the same. Details like difference between rpm or dpkg are not important. For learning all those choices are good.

I've read several post declaring that it was the existence of "step by step" online help/guides -- that specified particular versions, software and locations with links -- that enabled the poster to get ACS installed, learn what they needed to produce useful sites and succeed with ACS.

I think, for the community to thrive and grow -- with the enthusiasm regular infusions of new blood provides -- that there needs to be the same sort of step by step detailed guides for newbies now as there was when RH 6.2 was current.  One for each platform known to work be it RH  7.2, FreeBSD, OpenBSD Potato or whatever.  A guide someone new to *nix could sit down with and, assuming no incompatible hardware, follow to have a working Open/ACS system to learn and grow on.

While it's true that "you can just use whatever you like" and "lots of folks use RH 6.2 in production just fine", things are not quite so easy for a newbie.  Take RH 6.2 for instance.  The newbie will install the base distro -- then be faced with applying all the patches before they get rooted.  Quite a task for a newbie to even get to the point of being able to do useful ACS stuff.  For someone who has applied the patches over time, no big deal.  It is a big deal for a newbie starting from scratch with the base distro.  If there were a distro consisting of RH 6.2 with all patches applied and an up to date 2.2.20 kernel -- along with RPMs of the other packages needed for OpenACS, then "use this/our distro based on RH 6.2" would be OK advice for newbies IMO.  But such is not the case today AFAIK.

I wish I had the time and expertise to produce the above.  Because I would.  Lacking the existance of said newbie resources, I think it would be quite helpful -- and in the community's benefit -- for current documentation to be at least updated with an eye to current versions and platforms that work -- and preferably enhanced to be more like the step by step guides of old geared to newbies working through the home study course and problems sets.

It's these newbies that will grow into producing and enhancing new ACS packages/modules.  IMO, this community could really benefit from having OpenACS specific problem sets for working thru similar (or even the same, updated) learning task as existed in 1999 for Classic ACS, but dealing with Postgres instead of Oracle.  The old problem sets have links to resouces that no longer work.  Not good for newbies.

Again, I want to be able in good conscience to steer newbies to Open/ACS; but need to both know what specific advice to give as to what platform to get started with and know there are enough workable learning resources for them to sink their teeth into and learn how to develop useful communities and sites -- very similar to what one would learn working thru the Classic ACS problem sets in 1999.

Thank you for your input on this,

Louis

OS: RedHat 7.2 (because it's as good as anything else and most popular which means that you have the greatest chance to find the software/help from other people on the net).

If you're in Europe, I'd make that SuSE. At least at one point in time, they were outselling Windows in Germany.

I don't think that ACS/OpenACS is a good platform to learn. It's way too complex and learning curve is too steep. It's a good platform to build sites if you have a lot of experience and/or are ready to bite the bullet. I can imagine that a newbie might get to the point of launching the default site following step-by-step guide but:

• he won't learn much in the process
• he won't be able to do anything more than that (and become frustrated as a result)

While psets are useful, don't make it sound like nothing can be learned without them. They are just problems to solve with some tool specific hints about how to solve them. If one really has no idea what would be useful to do, it should be possible to solve pset problems in just about any web server that can embeed some scripting language and talk to a database.

Yon, I think you're going to have to support that claim about OpenACS not being a good platform to learn. Of all the platforms I've worked with, it's one platform where there is a lot less "voodoo magic" going on than on many other platforms.

I think, in general, learning how to build good db-backed web sites is tougher than most people have made it out to be. It's not as hard as, say, building an RDBMS, but it's harder than writing some HTML, too.
OpenACS is not perfect by any means, but I've found that most tools that are simpler to use than OpenACS are tools in which one builds systems that are not efficient, make poor use of the database, or later cause huge architectural problems.

But maybe you had something else in mind...

Hey Louis, I was in your position almost a year ago. I started off at ArsDigita.com, and came from a windowsish background. After reading Phillip and Alex's book, and finaly getting the ArsDigita pun (I'm not in to music or hisory or latin) I started to mess around with ACS (1999 was fairly recent then, and PGSQL wasn't in a state that I considered reliable enough for my goals).

Anyway, the point: By installing on various platforms with various walkthroughs I was able to learn only what was missing from X walkthrough at a time. Most I have learned about Linux, I have learned from trying to get Oracle to work on my RH-7.0 box.

If you want to learn SQL or run a site right away, then advanced walkthroghs are what you need, at least until somthing goes wrong. Like me sitting in new house with a Duron 750 and a Zoot CD, and not knowing how to properly recompile the kernel so it works with my Chip)

However in light of your post I have come to think that if the newbie community grows, It may be nice to make a newbie section in the discussion forum, and even some newbie docs on the web site.

Good luck Louis, and I hope I will be able to apply whatever knowledge I gain/have gained to helping you out with the OpenACS.

Ben, what I meant is that no complex platform (which includes, but is not limited to, ACS/OpenACS) is a good way to learn, at least not from the point of view of a newbie. It's because the combined complexity of *unix, managing AOLserver, managing postgres, knowing Tcl and SQL etc. is way beyond what I expect newbie to master in a short period of time. If someone doesn't know what upvar in Tcl does or how to write join in SQL then it's hard to expect that he'll understand a complex system like ACS.

What I would consider to be a good learning path is to first get reasonably good with SQL by playing with just the database, get reasonably good with Tcl/adp by playing with just AOLserver, then get a grasp on basics of HTTP like cookies and headers and only then start playing with systems like ACS that combine all of the above in a quite complex way.

I didn't mean that there is any other similiar toolkit that would be better but that a newbie should do things step by step because a complete toolkit (any reasonable featureful toolkit) is probably more than one can bite at a "newbie" level.

Agreed. In theory you could have a few kind-of-psets as described by Louis that would take someone from the point of having a vanilla OpenACS site running to "a few cool things that can be built with ACS fairly easily". If well done such document could be very useful for novices (for the reasons given by Louis). But such document is purely hypothetical beast at this time and "well done" is not easily achieved in practice.

I remember seeing this presentation on the aD webpage sometime ago.

Does anyone know what came of this?? Maybe we could borrow a few ideas.

Docs on how to get this "reference implementation" as they called it would go a long way to helping us newbie get up to speed with the ACS.

All the great things(games?) have an easy mode (chess, baseball, soccer) and almost anyone can take part. If you are really serious though, you can dig deeper (grandmaster, world cup etc). I think if we all think of the greatest games/tools, you could get involved in a couple of hours. If you were committed, you could eventually be miles ahead of the competition.

Striking this balance between ease of use and power will not even happen overnight, but is where we should be heading.

I have OpenACS running on OpenBSD (2.8 with a recent make world).
I should have the source and everything cleaned up in the next few days.  If you up for making and aolserver and openacs port for OpenBSD I'd be more than willing to help out.

Hey, Louis--you want a clue how to get started? Why don't you check out my "Advice for Bootcampers and Other Beginners" and "Problem Set Zero" and the like? After all, I basically wrote them for you.

The docs are all here.

While some of it is, no doubt, out of date, most of it should still be quite useful in orienting you. Enjoy.

Louis asked me to respond to this, so I'm respond...

The main things that computer programmers have to learn in order to become Internet application developers are (1) concurrency control (RDBMS programming and SQL), (2) data modeling (SQL again), and (3) page flow design (user experience).  The old 6.916 problem sets were an attempt to teach these things as quickly as possible and also introduce ACS, which the students would be using immediately afterwards as a substrate for their projects.  I still think that these psets are fairly effective on a per-student-hour basis.

If a student were setting everything up him or herself and wanted to learn the preceding concepts plus some of the details of ACS, I guess I'd recommend (a) Linux (I'm not qualified to say which version is best but I'd want one with a journaled file system), (b) PostgreSQL (easier to install and maintain than Oracle), (c) OpenACS 4.x (vibrant developer community of whom questions may be asked).

I think it may be a problem that nobody in the OpenACS world has picked up the old psets and edited them so that the step-by-step instructions work with PostgreSQL and OpenACS 4.x.  Learners always get stuck on the trivial stuff, e.g., finding the error log in the file system, rather than the slightly deeper stuff, e.g., interpreting the contents of that error log.

If you are concerned about security there is good book that guides you through how to set up an intranet using secured services (Apache-SSL, PostgreSQL), etc with Red Hat 6.2. The new version of this book is written for version 7.1/7.2 but they only released the book under GPL licence only after it has been for purchase for a while (hope that time will come soon :)). Hope this link helps

http://www.openna.com/products/books/securing-optimizing-linux/old.htm

For tcl, postgres, aolserver, linux etc. specific questions, it seems other sites specializing in those specific technologies best address them.  For example, the most obvious place to find answers to linux security issues is to discuss it at linux-centric websites.  OpenACS needs to focus on it's own technology and how it integrates with other technology.  With that it mind...

Philip Greenspun has identified my current problem: "PostgreSQL and OpenACS 4.x. Learners always get stuck on the trivial stuff, e.g., finding the error log in the file system, rather than the slightly deeper stuff, e.g., interpreting the contents of that error log."(Greenspun).

How does the OpenACS installation architecture differ between Oracle and Postgres installations?

More specifically, Where is the openACS installation architecture described for a postgreSQL-OpenACS4.x installation?

For the record, I have been RTFMing at https://openacs.org/doc/architecture-install.html and https://openacs.org/doc/installation.html documents as well as the others listed in this thread and other threads/forums... and references to 4.x documentation at the bottom of the 3.x docs at https://openacs.org/doc

Hello Torben,

Thank you for sharing your thoughts.  I would like to respond; not so much in the spirit of debate as just sharing my point of view for you and others in the community to consider.

Of course the RTFM approach *does* work.  However, I think there are several drawbacks for BOTH the newbie AND the community.

First, as I understand it, one of Mr. Greenspun's objectives with the "home study course" and problem sets was to expose the newbie to JUST the amount of *nix, database, scripting language, SQL and whatever else they needed *at that level of their expertise*, while stretching them and letting them learn how to find out more.

IIRC, I read something to the effect that Mr. Greenspun felt this was superior to the usual approach of spending a semester learning X, then Y, then Z then finally getting around to developing a web service.

With newbies, they don't know how much of what they need to learn at the early stages.  This is the main problem.

They risk getting side tracked for too long -- and in the wrong areas.  They also risk getting "lost" to the community since life is so busy and so many things in the open source world are competing for their attentions besides the need to RTFM X dozen times before they get to the point of having positive reinforcement for their efforts.

Lets say that by taking the RTFM approach, a newbie -- who keeps at it -- takes an extra 1-3 months to get productive.  Lets say in that time they could have gotten another site up for someone if they had better, up to date and security minded step by step guides for newbies when they started.

By the time 1000 newbies take this path, there would be another 1000 openACS sites out there providing POSITIVE PR for the community.  This is one of the things I had in mind when I talked about such up to date documents helping the community thrive.

Furthermore, lets say the newbie, being new to *nix -- let alone *nix security -- just takes the RTFM approach and "just does it".  They install RH 6.2 for instance.  Heck, that's what the docs say to use.  They even apply some patches.  At least the ones they THINK they need.

Not having been burnt yet, they don't really appreciate the need for great security yet.  Even if they did, they wouldn't know how to go about installing and configuring things securely.  THAT'S the point.  Sure, they can get there if they just RTFM -- eventually.  Eventually sure can be a long time for a newbie when it comes to *nix security.

1000 newbies do this, put up 1000 OpenACS sites that give the community a black eye in the minds of others as being software that's "not secure".  That's how suits and non geeks think.  Saying the newbie is at fault and didn't kow what they were doing would be moot at that point since 1000 potential paying customers who might have needed to hire OpenACS skilled people would have already decided not to use OpenACS anymore.  1000 organizations the newbie "helped" now swear off OpenACS.  They choose some other software for their web/net needs.  1000 potential jobs for someone with OpenACS expertise just went bye bye.

How different it might be if the newbie has an OpenACS "distro" / ISO that allows them to set up and offer OpenACS on top of up to date securly configured software foundation consisting of securely configured and installed Linux, or *BSD.  Because the newbie didn't start out with up to date securely set up software, the above IS a risk -- not just to the newbie but to the impression the public will develop of OpenACS.

Sure, there are very experienced OpenACS community members who would have no trouble setting up secure web service installations for folks.  But for any community to thrive, new blood and new members are needed.  This is what newbies are of course.  Why not help them not to bleed so much when they get started?  Why not do so by helping the newbie out with either step by step guides and/or some *nix and *BSD "distros"?

Sure it's not the fault of OpenACS software if the next 1000 newbies botch the web services they do for the next 1000 organizations.  It's the fault of poor system administration.  My point is, why not take the risk of the community getting BAD PR away by having up to date documentation that incorporates GOOD security practice/installation/configurations using up to date *nix like recent Linux, FreeBSD and OpenBSD versions?

This would help the community in getting GOOD PR in this post 9-11 more security conscious age by helping newbies set up web services using OpenACS based on secure implementations/configurations.

My thinking is it not only benefits the next 1000 newbies -- it benefits the next 1000 opportunities for the community to get GOOD instead of BAD PR in the minds of those making decisions on what software to avail themselves of -- and hire folks to provide and maintain.  Since OpenACS is not a monopoly, of course, this seems like an important consideration to me.

Basically, IMO, what's needed is something much like the old "home study course" and problem sets based on up to date software that community members having security expertise have "blessed".

Something a newbie can sit down with, follow and have at least all the software installed on a box they can hang off their internet connection and NOT get rooted or otherwise compromised.  Something they can then go on to install on a clients server and then "customize" for the clients needs.

THEN, thanks in part to the efforts of the next 1000 newbies armed with their OpenACS distros, we might see more and more demand for OpenACS expertise in addition to good PR that OpenACs not only does the job but gives the suits and non geeks "warm fuzzies" in the area of security -- a big selling point going forward, IMO (that's NOT going away).

I think basing the communities documentation off stuff like up to date Linux, FreeBSD and OpenBSD versions would help.  To me, telling some newbie to just RTFM and get a stock RH 6.2 distro and apply the patches (lots of lots if you haven't checked recently) then download bastile and follow that borders on the type of hazing some frat would do.  And if the newbie gets rooted and their OpenACS web services they do for some local organization for "gratis" since they are learning gets a black eye due to their newbie ideas of what constitues OK security, the community loses instead of gains.

Better to have an OpenACS distro of Linux, *BSD or whatever for the newbie to have a secure set up after instalation IMO.  This would benefit the community as well as the newbie.

The next 1000 newbies ARE going to wonder thru here.  Disregarding the benefit to the newbie, the question is how much benefit is the community going to derive from their 1000 journeys?  I think having up to date docs with up to date software with step by step instructions leading the newbie to at least have a secure setup after installation would provide a lot of good PR for the OpenACS community -- and eliminate the risk of unjustified blame and bad PR in the minds of others who make decisions on which software to use for their web services -- OpenACS or something else.

Best Wishes and Happy Holidays to everyone,

Louis

Hi Don and Louis,

Sorry for the delay... I forgot to click "Notify me of new messages" on this thread. [Is there a way to be notified of all messages all threads automatically? I like information overload]

Louis, your thoughts are right on the button of mine --exactly. I was just about to post a draft OACS infrastructure schematic, but thanks to Don's message, I'll contact the "dockeepers" and see if I can help accelerate the docs.

holiday cheers to you and everyone!

Actually, interface-wise, it makes sense for you to automatically receive email notifications for a thread you've posted on. Or at least to give the user that option explicitly. It's how I would customize the bboard to work.