[2741] in Release_7.7_team
Meeting Wednesday; AUI considerations
daemon@ATHENA.MIT.EDU (Greg Hudson)
Mon May 14 01:13:33 2001
Date: Mon, 14 May 2001 01:13:28 -0400
Message-Id: <200105140513.BAA04097@egyptian-gods.MIT.EDU>
From: Greg Hudson <ghudson@MIT.EDU>
To: release-team@mit.edu
Cc: aui@mit.edu
There will be a release team meeting this Wednesday at 1pm. Beta
should also happen around then, and probably will, although there are
some update safety issues (read: space checks) we ought to try to
resolve before then if at all possible, and some things like Sunblade
support which it would be nice to have.
I drafted the following note two days ago and decided to run it by
Bill and Andrew before sending it to the team. They thought it was
reasonable to raise the question but didn't agree with my answer. I
have some new comments at the end. I'd like to discuss this issue at
least briefly in release team, but I think the answer will be "press
on for beta, reevaluate before early," since we're not yet at the
point of no return.
---
I've been doing a lot of thinking lately about developments in IS and
how they relate to Athena, and I've concluded that I have to ask a
very difficult question, especially coming after we've all put so much
effort into this project:
Can we justify the cost of changing the default Athena user
interface under the current circumstances?
Here are the reasons why, as painful as it is to admit it, I currently
think the answer is no:
1. Athena development has been reduced to a skeleton crew. (A
conversation with rferrara today confirmed that there is no
possibility of replacing tb this fiscal year, and I can't
imagine being able to find new money for an Athena
developer given the current set of IS goals.) With five or
six EFT developers, we could maintain the environment while
making mildly ambitious improvements like AUI; with 3.5 EFT
developers, I think AUI is over the top.
The above statement may seem too conservative to some
people, especially with most of the AUI delivery work
already done. Consider that we have to be able to
accomodate the possibility of turnover and even further
headcount reduction, that we are doing a relatively poor
job of fixing problems in the Athena 8.4 environment, and
that a great many of the problems with the new user
interface will not be discovered until after it goes live.
2. The deliverables of AUI have been trimmed back to the point
where it is now basically a facelift. The only real new
value to users is that the new environment resembles
Windows more (but not in the way they care about,
i.e. being able to run Word) and is more customizable. The
menus have been reorganized, but we could trivially apply
that work to dash if we want to.
3. IS is currently committed to a direction wherein Athena
will no longer be the centerpiece of academic computing in
a few years (source is rferrara, again). Many of the
originally perceived benefits of AUI are predicated on the
assumption that AUI would become a framework for long-term
improvements (GUI versions of operations currently done on
the command line); it now appears that Athena does not have
enough of a future to warrant ever implementing those
improvements.
4. I am increasingly convinced that our testing strategy is
not adequate for this type of change. A majority of our
testers do not use the default user environment, so for the
most part AUI would receive only cursory testing before
going live.
5. Overcommitting ourselves with AUI may make it more
difficult for Athena staffers to explore roles in new IS
projects, which seems unfair if Athena development is going
to be an eventual dead end.
If we do not change the default interface, I see no reason why we
should not leave the GNOME software in the release, and even allow
people to use the new environment by touching
~/.athena_gnome_interface. But it wouldn't have to be documented by
TPS, and it wouldn't be as critical for dev to fix irregularities in
the GNOME interface or for OLC to be able to answer questions about
it. And, of course, it would not be forced onto existing users who
have already learned the current Athena interface.
Please think seriously about this question; we will discuss it at
release team on Wednesday. (People are also welcome to reply to this
message, of course.)
---
Whether Athena is really a dead end is still somewhat in question. I
know that IS will be submitting a proposal to Curry for new money for
a scalable 1-1 computing project, and it's conceivable (although I
have doubts) that some of that money could be spent on staff to adapt
Athena for the Linux side of these laptops. Andrew pointed out that
having AUI might be good ammunition in fighting for more resources of
this nature.
I remain concerned that if Athena does turn out to be a dead end, we
could be putting ourselves in an untenable position for maintaining it
during the transition. It will be a hell of a lot easier to keep
Athena going on a skeleton crew if there isn't a big festering pile of
GNOME code in the critical path between users and their work.