Re: Known MacOS programmer about WO/EOF

[David Ayers]

Armando Di Cianno wrote:

> Take for example gsweb.  Almost 100% undocumented, never been
> "released" (come on -- it's at least 0.3, no?) and yet, if you try it
> out, it's extremely cool.  So what on earth do we tell
> end-user/developers?  "You know...try it out...and if you happen to
> have developed for WO about 5 years ago, it's almost feasible to
> use...otherwise...well...you'll figure it out....".  Pretty much, not
> cool.

OK, I'll take the bait...  To me, making a release involves a certain 
responsibility wrt user expectations.  A certain level of feature 
completeness and usability.  Since you have setup gsweb before I'm sure 
you know all the extra twiddling needed.  Things like the Apache adaptor 
module and it's configuration.  In fact I'm currently working on 
./configure for the Apache adaptor right now so some of that will be 
remedied.

But there are more user expectations.  For one thing I'm not planing on 
supporting backward compatibility to any current NSCoding 
implementations in GDL2 or GSWeb.

Also there are still many issues when you your users expect the high 
level EOF/WebObjects API and the find out that they have to work around 
many incomplete low-level implementations.  Yes GSWeb and GDL2 can be 
used by people who already know how it is supposed to work and fill in 
holes where necessary.  But making a release says to me "Users, you may 
start building apps on top of this and we will support you".  I can't do 
that.  Most of my time is spent building on top of what we currently 
have, but I try to clean up and write API tests whenever I can.  I once 
started auditing the what needs to be done for GDL2, but decided my 
limited time is better spend in just getting things done and 
coordinating things with other developers on a case by case basis.

What we need is developers that already know how EOF/WebObjects and can 
spend time working on it.  There are a few that are trying to get into 
it.  But they often lack the experience and resources (reference 
implementations) to really get the correct implementation.  I'm sad to 
say, that I simply don't always have the time or even the know-how to 
help them out, even though I try when I can.  Another problem I see, is 
that some of the work to squeeze out more efficiency has made the code 
much harder to read, making it even harder for new folks to help out. 
I'm still hoping to find a way to remedy that, but to do that I'll need 
to find some time to do some real benchmarking myself to make sure we 
don't loose measurable performance.

Cheers,
David