• Paul D. Anderson <paul.d.removethis.anderson comcast.andthis.net> Jun 24 2009
• Jason House <jason.james.house gmail.com> Jun 24 2009
• Paul D. Anderson <paul.d.removethis.anderson comcast.andthis.net> Jun 24 2009
• Tim Matthews <tim.matthews7 gmail.com> Jun 24 2009
• Jason House <jason.james.house gmail.com> Jun 24 2009
• Tim Matthews <tim.matthews7 gmail.com> Jun 24 2009
• Stewart Gordon <smjg_1998 yahoo.com> Jun 25 2009
• Paul D. Anderson <paul.d.removethis.anderson comcast.andthis.net> Jun 25 2009
• BCS <ao pathlink.com> Jun 25 2009
• Paul D. Anderson <paul.d.removethis.anderson comcast.andthis.net> Jun 25 2009
• Ary Borenszweig <ary esperanto.org.ar> Jun 25 2009

Paul D. Anderson <paul.d.removethis.anderson comcast.andthis.net> writes:

A. What I'm looking forward to in the near future:

1) D2 will be frozen.
2) Andre's book will come out.

B. What I'm hopeful for, and really expect to see fairly soon thereafter:

3) A complete, consistent D language specification (for D1 and D2).
4) D2 support for many of the popular tools and libraries out there, especially
Tango and descent.
5) Continued development of D compilers alongside DMD: LLDC, GDC, dil, etc.
6) Expansion of Phobos with more contributions from others besides Walter and
Andre.
7) Improvements in the D toolchain, as have been discussed here at some length:

Quoting Frank Rundell:
"If it wants to compete with the 'big boys' it needs an IDE, a GUI library that
can compile and work, a debugger that understands D, a proper linker, packaged
releases for linux, an installer for Windows, etc."

C. What I'd like to see in the longer run:

8) A solid suite of D tools, written in D. 

D. The good news is that none of these except 1) and 2) depend on Walter or
Andre. All they need is a some committed effort from a group of intelligent,
knowledgeable, capable, interested supporters. Hmmm......

E. I was going to add some helpful advice (i.e. ranting) about working together
and taking the bull by the horns and running it up the flagpole to see if the
cat licks it up, etc. but I'll refrain. I certainly don't want to denigrate all
the effort that has been made -- I just want to see more of it! :)

F. Note that I've identified all my comments by helpful numbers and letters so
you can reply:

"Gee, Paul, I thought point B4 was particularly enlightening. Can I send you
some money?"

Paul

Jason House <jason.james.house gmail.com> writes:

What happens when #3 turns up an inconsistency in the language spec after #1
and #2 have occurred? It'd be a real shame to have to wait another 2-3 years
for it to be corrected in a stable DMD!

If #4 hits issues, #1 may block resolution of those issues.

#6 won't occur after #1 either. If history serves, a "stable D" translates to
no Phobos enhancements. The world may require a D2 Tango that replaces Phobos
once again.


I honestly believe that we have to start the D2 integration now in order to
ensure changes to #1 and #2 occur. If we wait and issues crop up, there are
only two possible outcomes:
• Having permanently bad features
• A mismatch between Andrei's book and the D2 spec.

   

Paul D. Anderson Wrote:

 A. What I'm looking forward to in the near future:
 
 1) D2 will be frozen.
 2) Andre's book will come out.
 
 B. What I'm hopeful for, and really expect to see fairly soon thereafter:
 
 3) A complete, consistent D language specification (for D1 and D2).
 4) D2 support for many of the popular tools and libraries out there,
especially Tango and descent.
 5) Continued development of D compilers alongside DMD: LLDC, GDC, dil, etc.
 6) Expansion of Phobos with more contributions from others besides Walter and
Andre.
 7) Improvements in the D toolchain, as have been discussed here at some length:
 
 Quoting Frank Rundell:
 "If it wants to compete with the 'big boys' it needs an IDE, a GUI library
that can compile and work, a debugger that understands D, a proper linker,
packaged releases for linux, an installer for Windows, etc."
 
 C. What I'd like to see in the longer run:
 
 8) A solid suite of D tools, written in D. 
 
 D. The good news is that none of these except 1) and 2) depend on Walter or
Andre. All they need is a some committed effort from a group of intelligent,
knowledgeable, capable, interested supporters. Hmmm......
 
 E. I was going to add some helpful advice (i.e. ranting) about working
together and taking the bull by the horns and running it up the flagpole to see
if the cat licks it up, etc. but I'll refrain. I certainly don't want to
denigrate all the effort that has been made -- I just want to see more of it! :)
 
 F. Note that I've identified all my comments by helpful numbers and letters so
you can reply:
 
 "Gee, Paul, I thought point B4 was particularly enlightening. Can I send you
some money?"
 
 Paul
 

Paul D. Anderson <paul.d.removethis.anderson comcast.andthis.net> writes:

Jason House Wrote:

 What happens when #3 turns up an inconsistency in the language spec after #1
and #2 have occurred? It'd be a real shame to have to wait another 2-3 years
for it to be corrected in a stable DMD!
 
 If #4 hits issues, #1 may block resolution of those issues.
 
 #6 won't occur after #1 either. If history serves, a "stable D" translates to
no Phobos enhancements. The world may require a D2 Tango that replaces Phobos
once again.
 
 
 I honestly believe that we have to start the D2 integration now in order to
ensure changes to #1 and #2 occur. If we wait and issues crop up, there are
only two possible outcomes:
 • Having permanently bad features
 • A mismatch between Andrei's book and the D2 spec.
 
    


I agree with you on all but the point about #6. I'll admit that that has been
the case but I don't see it as defining futue behavior. Especially if the
changes come from other than Walter/Andre.

But your point is well taken. We definitely want to catch breaking changes
before the freeze. I was making the (Pollyannaish? Panglossian?) foolish
assumption that D2 would be beyond reproach. Nu-huh.

Paul
 
 Paul D. Anderson Wrote:
 
 A. What I'm looking forward to in the near future:
 
 1) D2 will be frozen.
 2) Andre's book will come out.
 
 B. What I'm hopeful for, and really expect to see fairly soon thereafter:
 
 3) A complete, consistent D language specification (for D1 and D2).
 4) D2 support for many of the popular tools and libraries out there,
especially Tango and descent.
 5) Continued development of D compilers alongside DMD: LLDC, GDC, dil, etc.
 6) Expansion of Phobos with more contributions from others besides Walter and
Andre.
 7) Improvements in the D toolchain, as have been discussed here at some length:
 
 Quoting Frank Rundell:
 "If it wants to compete with the 'big boys' it needs an IDE, a GUI library
that can compile and work, a debugger that understands D, a proper linker,
packaged releases for linux, an installer for Windows, etc."
 
 C. What I'd like to see in the longer run:
 
 8) A solid suite of D tools, written in D. 
 
 D. The good news is that none of these except 1) and 2) depend on Walter or
Andre. All they need is a some committed effort from a group of intelligent,
knowledgeable, capable, interested supporters. Hmmm......
 
 E. I was going to add some helpful advice (i.e. ranting) about working
together and taking the bull by the horns and running it up the flagpole to see
if the cat licks it up, etc. but I'll refrain. I certainly don't want to
denigrate all the effort that has been made -- I just want to see more of it! :)
 
 F. Note that I've identified all my comments by helpful numbers and letters so
you can reply:
 
 "Gee, Paul, I thought point B4 was particularly enlightening. Can I send you
some money?"
 
 Paul
 

Tim Matthews <tim.matthews7 gmail.com> writes:

Paul D. Anderson wrote:
 A. What I'm looking forward to in the near future:
 
 1) D2 will be frozen.




I am really for having Bartosz Milewski's changes implemented in D2 
rather than d3, or further. Multi core cpus are here now and it would 
help D get noticed as well as benefiting existing D devs.

Jason House <jason.james.house gmail.com> writes:

Tim Matthews Wrote:

 Paul D. Anderson wrote:
 A. What I'm looking forward to in the near future:
 
 1) D2 will be frozen.


 
 
 I am really for having Bartosz Milewski's changes implemented in D2 
 rather than d3, or further. Multi core cpus are here now and it would 
 help D get noticed as well as benefiting existing D devs.


There is still debate over exactly what will go into D2. Which features of
Bartosz's design do you like? Dislike? Some feedback may help the offline
discussion.

Personally, I like 
• the automatic ownership of member variables
• locking requirements are known at compile-time
• explicit lock free
• scope parameters (AKA lent)
• unique as a type

I dislike:
• completely automated locking (may lead to recklessness)
• verbose notation

Tim Matthews <tim.matthews7 gmail.com> writes:

Jason House wrote:

 There is still debate over exactly what will go into D2. Which features of
Bartosz's design do you like? Dislike? Some feedback may help the offline
discussion.


I would need to have a long think about this. I think a language that is 
designed for nothing but parallelism may benefit a lot and D could do 
with some as a next generation language.

s ~= " ";

Here the syntax looks atomic but the semantics is not because of the 
length property. I was kind of thinking I liked this locking, then 
thought maybe too much overhead or verbose when I didn't need the lock.

I think everyone can agree on a dislike for a verbose notation so I hope 
  D can make the most of angle brackets and colons to minimize this.

Stewart Gordon <smjg_1998 yahoo.com> writes:

Paul D. Anderson wrote:
 A. What I'm looking forward to in the near future:
 
 1) D2 will be frozen.
 2) Andre's book will come out.
 
 B. What I'm hopeful for, and really expect to see fairly soon thereafter:
 
 3) A complete, consistent D language specification (for D1 and D2).



Uh, the sequence of items 1 and 3 makes no sense at all.

Stewart.

Paul D. Anderson <paul.d.removethis.anderson comcast.andthis.net> writes:

Stewart Gordon Wrote:

 Paul D. Anderson wrote:
 A. What I'm looking forward to in the near future:
 
 1) D2 will be frozen.
 2) Andre's book will come out.
 
 B. What I'm hopeful for, and really expect to see fairly soon thereafter:
 
 3) A complete, consistent D language specification (for D1 and D2).


 
 Uh, the sequence of items 1 and 3 makes no sense at all.
 
 Stewart.


The specification that we have in the digital mars web pages is incomplete and
inconsistent. It has always lagged the compiler development. You are correct
that this is nonsensical, but it is what it is. (Part of the reason is that D2
is intentionally experimental -- Walter tries ideas and pulls them in or out as
the language develops, so he can't write the "specification" beforehand --
usually he updates the web pages when new language features are made public.)

I'd like to see a more formal specification, even if it is ex post facto. I
doubt Walter will be providing that. I'm hoping someone steps up to make it
happen. I'm willing to do it myself, but I need to put the finishing touches on
the bigfloat implementation. When that's done, and if no one else jumps in,
I'll tackle it.

Of course there's no reason why such a "snapshot" specification has to wait
until the design is final. It can be based on any current snapshot and updated
as changes occur. Again -- 100% backwards, but better, IMHO, than not done at
all.

Paul

BCS <ao pathlink.com> writes:

Reply to Paul,

 Of course there's no reason why such a "snapshot" specification has to
 wait until the design is final. It can be based on any current
 snapshot and updated as changes occur. Again -- 100% backwards, but
 better, IMHO, than not done at all.
 


This makes me think of how to maintain it. What would be cool is if each 
spec item had an attached blob of code that checks and demonstrates the
assertion 
made by the item. Then by switching to the latest and greatest and just
re-compiling 
all the blobs you could get a list of items that need to be rewritten. What 
would be very cool would be all of that in a wiki format that automatically 
marks out of date sections.

Paul D. Anderson <paul.d.removethis.anderson comcast.andthis.net> writes:

BCS Wrote:

 Reply to Paul,
 
 Of course there's no reason why such a "snapshot" specification has to
 wait until the design is final. It can be based on any current
 snapshot and updated as changes occur. Again -- 100% backwards, but
 better, IMHO, than not done at all.
 


 This makes me think of how to maintain it. What would be cool is if each 
 spec item had an attached blob of code that checks and demonstrates the
assertion 
 made by the item. Then by switching to the latest and greatest and just
re-compiling 
 all the blobs you could get a list of items that need to be rewritten. What 
 would be very cool would be all of that in a wiki format that automatically 
 marks out of date sections.
 
 


Good idea. Now if we can get someone to bell the cat....

Ary Borenszweig <ary esperanto.org.ar> writes:

Paul D. Anderson wrote:
 A. What I'm looking forward to in the near future:
 4) D2 support for many of the popular tools and libraries out there,
especially Tango and descent.


I'm finishing updating Descent to DMD 1.045. Then I'll update it to also 
support the latest D2. But I'm doing this very slowly.