Status
Gavin King
1. Fix high priority bugs before xmas. (We have 3 weeks.)
2. Take a bit of a break for 2 weeks.
3. Release M1 in the second week of Jan.
The high-priority bugs are:
## Compiler
All of these, and perhaps we'll run into some additional ones along the way:
## IDE
Perhaps a couple of these:
(#102 looks like a good one to target.)
## Language module
This one:
(I have a little more cleaning up to do on the language module, but
most of the work is now done and tested.)
## Typechecher
These two:
--
Gavin King
gavin...@gmail.com
http://in.relation.to/Bloggers/Gavin
http://hibernate.org
http://seamframework.org
Gavin King
* support for module versioning
* the module runtime
* default arguments
That kinda hurts, and so I propose that we follow up with an M1.1
release that has support for these three things.
I further propose that we split M2 into two separate milestones:
M2A: higher-order functions and switch/case
M2B: mixin inheritance, member classes, and type aliases
i.e. M2A emphasizes functional programming, M2B emphasizes
object-oriented programming.
Stephane Epardaud
Note that this plan means that the following will be missing from M1:
* support for module versioning
* the module runtime
* default arguments
Stephane Epardaud
Gavin King
OK, fine, so here's a new roadmap proposal that divides the work into M1-M5:
http://ceylon-lang.org/documentation/roadmap/
The main reason I had been avoiding this is that I wanted the Ms to be
of approximately the same "size". Now that's not true, so someone
might think, "well, it took them a year to get M1 out, so M5 will be
due in late 2015". ;-) Actually, I think each of M2-M5 is an
approximately two-month iteration.
But folks, doing a release once every two months is going to keep us
on our toes. We're all going to need to be working very consistently
if this is the plan. Otherwise, slippage of one little thing can throw
off the whole programme, since within each iteration there is much
less possibility for parallelization.
Tako Schotanus
OK, fine, so here's a new roadmap proposal that divides the work into M1-M5:On Thu, Dec 1, 2011 at 1:14 PM, Stephane Epardaud
<stephane...@gmail.com> wrote:
> Also, can we stick to MX format? M1.1 and M2A doesn't strike me as very nice
> ;)
http://ceylon-lang.org/documentation/roadmap/
The main reason I had been avoiding this is that I wanted the Ms to be
of approximately the same "size". Now that's not true, so someone
might think, "well, it took them a year to get M1 out, so M5 will be
due in late 2015". ;-) Actually, I think each of M2-M5 is an
approximately two-month iteration.
But folks, doing a release once every two months is going to keep us
on our toes. We're all going to need to be working very consistently
if this is the plan. Otherwise, slippage of one little thing can throw
off the whole programme, since within each iteration there is much
less possibility for parallelization.
Stephane Epardaud
Tom Bentley
* `switch`ing on Strings will make out own `switch` implementation easier
* The 'suppressed exception' feature will mean we can do something reaonably uncontentious with ceylon/ceylon-spec#66
* I think we should properly investigate using `invokedynamic` for named argument method calls. Time we invest further in the solution we have now will become wasted effort.
My circumstances mean I can't guarantee to work *consistently* on Ceylon: It's what happens in my spare time. I'll work as much as I can on it, but that doesn't correlate to anything concrete. It would be wrong of me to suggest a level of commitment that I can't actually fulfil.
Tom
Gavin King
release M3 in early May at the latest. That's two more releases in the
first half of the year.
--
Gavin King
Of course. I'm more referring to those of us who are on this project fulltime.