On Mon, 4 Jun 2001 [EMAIL PROTECTED] wrote:
> However, as we approach the 2nd half of 2001, one might ask whether
> there is an end in site, a point at which 1.1.6 will be intended as a
> release worthy to succeed 1.1.5fix in all respects? Has there been
> any loss of momentum? Is it possible that this is becoming another
> development path that has bitten off more than it can chew? Just my 2
> cents.
I think you are partially right. The development cycle up to 1.2.0 seems
too ambitious in retrospect.
It's always a balance: Do we want to encourage work by letting it
proceed in the main cvs, or should it have less attention by being
shoved away in a branch that won't affect the stability of the rest?
We can see the difference with the math code and the natbib nranch: I
think the math clean up is proceeding much more rapid in the main cvs than
it would in a separate branch, simply because it gets much more exposure,
and people file lots of useful bug reports. Thus Andre gets the feedback
he needs, and the code is maturing relatively quickly. Remember that the
math code has been dead for a few years, and it's in everybody's best
interest that the code is brought back to live. It would have been hard to
say to Andre at the start of the year: No, we don't want the math clean
up, given that it's very rare that anybody touches it.
Compare also with the natbib work that lives on a branch: I don't think
Angus has gotten much feedback on that. This in turns results in less
entuisiasm from the developer, and there is a real fear that the work will
never be merged. This is very unfortunate, because it both affects the
functionality of LyX, and what is worse: It might disencourage Angus from
doing more of the very valuable work, we all need!
So, what to do about it?
In retrospect, the problem with the 1.1.6 work is simply that too many
things have been done at the same time. Either the math rewrite or the
inset stuff should have waited until the other was good enough for a
stable release. However, there were many good reasons to do either at the
point in time, the work started: The inset stuff had to be done in order
to stabilize the table rewrite introduced in 1.1.6. And as mentioned, it
would have been bad for morale to turn down the math effort from Andre.
Also, at the time, it seemed that there would be a fair chance of managing
to the math work incrementally, such that the main stability would not be
affected. Unfortunately, it turned out that this was not possible. That's
just bad luck, and an indication that the math clean up is well due.
Notice also that the GUII work that has been going on is not as
disruptive. It's fairly seldom that work on the GUII stuff affects the
main stability, so there is no real problem with the GUII work proceeding
at the same time as the other core stuff.
Now, I would not compare the situation to the old development branch that
had to be completely scrapped. Yes, at the moment the math code is not
good enough for a stable release, but we are not several months or years
away as we was at that point. It should hopefully be a matter of a few
weeks of work.
Also, we have the emergency exit: Revert the math code back to the 1.1.6
stuff.
But there is good reason to believe that the necessary work can be done at
the next LyX developers meeting. So, I don't think the prospects are that
bad for LyX.
What we need now right now is more testing, and more bug reports. When we
are in Italy, we can choose between three different activities:
1) Drink beer
2) Fix bugs
3) Build new features
Obviously 1) and 3) are the most fun, so we'll for sure do some of that.
However, we also know that 2) is vital for the project to be succesful,
so we will do that also. But we can only do that if we have the bug
reports! So help us out with this. Come to Italy and work as a tester
together with us, and/or start now at home. Check out the cvs version and
file all those bug reports at us. Remember: Bug reports are as valuable as
code.
Thank you,
Asger
