Forum OpenACS Development: Scope & Timeline

Posted by Ben Adida on
The main work to be undertaken by the core OpenACS team will be in porting the ACS kernel and core. This will set the stage for various teams to port packages independently. will provide a home for all packages, but in the spirit of open-source, the development will remain decentralized to allow a maximal amount of user participation.

In terms of timeline, all people interested in discussing the design of OpenACS 4.0 should immediately download ACS Classic 4.0 alpha 2. Beta is coming soon, but you can gleam a lot of information from alpha 2 as is. The schedule will work as follows:

  • Week of October 1: all issues on the table, group discussions.
  • October 8-22: resolution of all issues, finalized discussions, task assignments
  • October 22-29: massive porting effort. Kernel and Core are tight pieces of code, thus a short-and-sweet focused effort will work best.
Remember, this is a discussion forum, so add comments and questions below! These are all proposals.
Posted by Dan Wickstrom on
Has anybody had any success loading acs classic 4.0 alpha 2. I get the following error while when running the installation:
[28/Sep/2000:23:06:23][8795.4101][-conn0-] Notice: No errors.

[28/Sep/2000:23:06:24][8795.4101][-conn0-] Notice: 	check (min_n_values <= max_n_values)

[28/Sep/2000:23:06:24][8795.4101][-conn0-] Notice: 	                       *

[28/Sep/2000:23:06:24][8795.4101][-conn0-] Notice: ERROR at line 26:

[28/Sep/2000:23:06:24][8795.4101][-conn0-] Notice: ORA-01450: maximum key length (1578) exceeded

[28/Sep/2000:23:06:24][8795.4101][-conn0-] Notice: comment on table apm_parameters is '

This appears to be happening at line 589 in apm-create.sql which corresponds to this section of code:
create table apm_parameters (
	parameter_id		constraint apm_parameters_fk 
				references acs_objects(object_id)
			        constraint apm_parameters_pk primary key,
	package_key		varchar(100)
				constraint apm_pack_param_pack_key_nn not null 	
				constraint apm_pack_param_type_fk
			        references apm_package_types (package_key),
	parameter_name		varchar(100) 
				constraint apm_pack_params_name_nn not null,
        description		varchar(2000),
	section_name		varchar(2000),
	datatype	        varchar(100) not null
			        constraint apm_parameter_datatype_ck 
				check(datatype in ('number', 'string')),
	default_value		varchar(4000),
	min_n_values		integer default 1 not null
			        constraint apm_paramters_min_n_ck
			        check (min_n_values >= 0),
	max_n_values		integer default 1 not null
			        constraint apm_paramters_max_n_ck
			        check (max_n_values >= 0),
	constraint apm_paramters_attr_name_un
	unique (section_name, parameter_name, package_key),
	constraint apm_paramters_n_values_ck
	check (min_n_values <= max_n_values)

I don't see anything unusual here, and it doesn't seem like a configuration error.
Posted by Dan Wickstrom on
I checked the log comments for apm-create.sql, and it appears that section_name was shortened from 2000 to 200 after the alpha 2 release.  The error message must be referring to the unique constaint which creates an index on section_name, parameter_name, and package_key, and the index exceeds the maximum key length of 1578. I'm going to try checking out one of their stable branches and see if that works any better.
Posted by Dan Wickstrom on
I checked out the revision lableled as "stable" and the installation went o.k.  I have to say that my initial impression was quite good.  They have pretty much automated the installation process, and once you get past that minor oracle installation task :), everything flows pretty smoothly.
Posted by Bob OConnor on

Ben wrote:
"...all people interested in discussing the design of OpenACS 4.0 should immediately download ACS Classic 4.0 alpha 2...

Well, I'm interested in discussing the design...but I don't have Oracle and space....

Is there a development machine somewhere where interested people can get a look at the "RUNNING" program from an outside level? One could then download the code to see how it works and make further comments without doing the install.


Posted by Ben Adida on
ArsDigita folks, do you have a server running with the latest ACS,
like Bob is proposing?
Posted by Nathan Reeves on
The main acs4.0 page mentions that there is a copy of ACS 4.0 running on this server which is rebuilt twice daily. As it stands I cannot get to the box, but I suspect that it's the firewall I'm going through that is preventing connection as opposed to the server not actually running.

Posted by Don Baccus on
I can't get into it, either, at least at the moment.  Perhaps they moved it to a different URL?
Posted by Gilbert Wong on
I have 4.0a2 running on a development machine at home but I can load it up on my public server if there is enough interest.  Let me know.
Posted by Gilbert Wong on
One more thing, is there more documentation on acs4.0?  I flipped through the docs in 4.0a2 and I didn't completely understand the new structure of the ACS (packages, parties, objects, etc.).
Posted by Bryan Quinn on
Greetings, I have made the ACS 4 Clean Server operational and it is now open as an ACS 4.0 sandbox. Please login to the system as user openacs password openacs. If anyone wants me to change these defaults, please let me know via email. The server is rebuilt once a night, so no changes will persist. Also, this sandbox does not let you modify code, but simply enables experimentation with a running ACS 4.0 system. If you happen to discover a bug when using the system, please report it using the ArsDigita SDM. Many thanks.
Posted by Don Baccus on
Thanks, Brian, this should help a lot.  It's GOOD that changes aren't allowed, as we'll be wanting an official, clean, undiddled ACS 4.0 Classic to compare with when we find bugs in our port.  And I'm certain that we'll be quick to pass along bugs (and hopefully bug fixes) as we stumble and mumble our way forward.
Posted by Roberto Mello on
I poked around the sandbox. It looks very nice. Thanks for that Bryan.