Forum OpenACS Q&A: OpenACS 4.x - what should "x" be?

Collapse
Posted by Don Baccus on
In the course of some private e-mail discussion, the suggestion was
made to call our first release "OpenACS 4.4" or "OpenACS 4.5".  I'm
more or less convinced that "> 4.2" makes sense but thought I'd post
here to get some feedback.

The rationale as presented to me is that

1. We've included elements from what would've been aD's ACS 4.3
release if they'd not dropped Tcl support (workflow, for instance), so
"4.2" is in reality inaccurate.

2. Along with bug fixes, we've added considerable functionality.
Dual-RDBMS support for starters (it's hard to underestimate that item,
given that it's eaten programmer-months of volunteer time!).  We've
added some new packages (ecommerce3, RSS, cronjob, a new search package
which at the moment is PG-only, etc).  A step ahead of what would've
been aD's ACS 4.3, in other words.

3. 4.4 vs. 4.5 - "4.5 is a nice round number, why not just start there?"

I think the argument for releasing as OpenACS 4.4 or OpenACS 4.5 (as
opposed to 4.2 or 4.3) is quite reasonable.  In regard to 4.4 or 4.5 I
don't really care.  Our next release will include a major cleanup and
dotLRN elements (in particular the new portal pacakge) - if we clean
up the core to the point where we really like it calling it "5.0"
seems likely (in part to make it clear that an upgrade from 4.x->5.0
likely will involve running some heavy-duty upgrade scripts, more work
than one would expect from a "+0.1" release).

So chime in, folks.  If necessary we could run another poll but I'm
hopeful it's not necessary.

Collapse
Posted by C. R. Oldham on
4.5
Collapse
Posted by MaineBob OConnor on

I bet a poll would work better. I tried to post my answer I got:

    "We had a problem processing your entry: There are already 1 messages in the database with the same subject line and body. Perhaps you already posted this? Here are the messages:
    2002-02-22 12:02:17-05 by C. R. Oldham (cro@nca.asu.edu) If you are sure that you also want to post this message, then back up and change at least one character in the subject or message area, then resubmit.

So I choose 4.5

-Bob

Collapse
Posted by Don Baccus on
That's pretty funny!  I don't think I've ever seen two replies that are exactly the same that weren't caused by a double-click.

As far as polling goes I don't want to take the time if possible.  It's not a moral opposition, just a practical opposition.  This just came up in mail the day before yesterday, before that I'd just assumed we'd call it "4.2".

I know at least four other folks who like "4.5" other than you two and me (i.e. those who got involved in our private e-mail discussion), so that makes seven "yeas" and no "nays" thus far.

Collapse
Posted by Titi Ala'ilima on
One thing to keep in mind is the presumably upcoming migration to Oracle9i compatibility.  This is going to present a major break in continuity, as there will be almost no direct compatibility between the two versions.  Granted, migrating will be relatively easy, but not quite automatic, so all the APMs will have to change.  We've decided to re-brand the version we've put out for 9i as M(odified)ACS, so that there will be no confusion about interoperability.  It will hopefully help us keep clear what should work with what.

OpenACS will also run into the same problem at some point.  Though that might take place at the 5.0 marker, so it'd be easier to make a compatibility break.

All that said, I'd probably go with 4.3 or 4.4.  I think the move from ACS to OpenACS is the biggest differentiator, and should be upheld in its own right.  Making a big numeric jump for a characteristically differenct product seems too much like aD's Java 4.6 fiasco.  So I'd say understate the version leap, but emphasize the essential technology leap in going from ACS Classic to OpenACS.

Collapse
Posted by Don Baccus on
Yes, we'll be tackling 9i for the 5.0 mark (if that's what we call it).  I'll also propose we only support PG 7.2, not 7.1, with that release.

So we'll have lots of discontinuity to go around, plenty for everyone.

The Oracle version will be the worst, the package stuff just requires replacing existing packages in your active install and we can hack packages to use the new names.  But ... anyone who's done customization will have to hack their own code to use the renamed functions and procs.

Ugh.

Collapse
Posted by Titi Ala'ilima on
Using the scripts we've provided, it shouldn't be hard to "hack" even customized packages.
Collapse
Posted by Ben Adida on
We could also provide parallel script for Oracle9 temporarily...
With all the security bugs in Oracle9, I don't suspect that people
will move off Oracle8 anytime soon. Any new code can be the
new Oracle 9 standard, but old code can simply stay as is, with
one Oracle8 version for it, and an Oracle9 version too. This will
massively smooth out the upgrade problem for OpenACS. We've
got a query dispatcher, let's use it!

Thoughts?

Collapse
Posted by Don Baccus on
No, but it won't be totally automated.  My thinking is that folks would expect a "+0.1" release to truly be incremental, i.e. offer a painless upgrade path, that's all.  Calling the next version "5.0" would make it clear that we're talking about a fairly major upgrade.

But that's kinda offtopic at the moment, something to worry about in a few months, not now.