Long awaited Demetra roadmap

Hello Robert,

I don't think that I hijacked it ... I commented on the roadmap, that's
all. Don't worry, if aspects are not even in a roadmap for 6.0 then I'll
have to say goodbye to IDEA anyway ... after 4 years. Sad but true ...

Hello Michael,

MR> Hi,
MR> MR> so you think that Aspects are not important? IMO they're the single
MR> most important new technological concept since object oriented
MR> programming.

Not again! Come on man, let's not hijack the thread.

R

+so you think that Aspects are not important? IMO they're the single most
important new technological concept since object oriented programming.+

Aspect-oriented programming will probably make the big-time some day, but the current batch of aspect-oriented languages and frameworks almost certainly won't. The programming models are too complex and brittle (AspectJ, in particular), the implementations too fragile, the tooling too weak, and their backers way too annoying. It's not enough to say IDEA should support aspect-oriented programming. You have to say just what flavor of aspect-oriented programming should be supported, and why it should be chosen over all the others. I don't see anything like an reasonable choice to make among the current contenders.

--Dave Griffith

VM> Server integration - any additional licenses required for the server
VM> side pieces? Assuming rather easy integration with CruiseControl?

It replaces, not integrates with CruiseControl. We haven't made any
final decisions on licensing yet.

Interesting.
Many companies have home-grown solutions "a la CruiseControl", so it would be interesting
to see if migrations to the IntelliJ solution would be easy.

I suppose those server parts in IntelliJ should be strongly documented(not like the OpenAPI),
otherwise no one will "change the winning team" :).

Thanks in advance,

Ahmed.

Michael Riess wrote on 17/11/05 23:57:

so you think that Aspects are not important? IMO they're the single most
important new technological concept since object oriented programming.

AOP is important, supporting AspectJ -ONLY- at the core is not.

In article <488e3f9b1151c8c7b8fcd043f740@news.jetbrains.com>,
max@jetbrains.com says...

can be found at http://www.jetbrains.net/confluence/display/IDEADEV/Demetra+Roadmap

------------------
Maxim Shafirov
JetBrains, Inc
http://www.jetbrains.com
"Develop with pleasure!"

First class Spring support, that is being about to
- recognized Spring config files,
- direct refactoring of config files
- ability to modify config files when classes are changed
- intention support for config files(like changes to make the files
less verbose
- Graphical display of config files
- able to detect missing refs
--
-

David H. McCoy


-

In article <488e3f9b1151c8c7b8fcd043f740@news.jetbrains.com>,
max@jetbrains.com says...

can be found at http://www.jetbrains.net/confluence/display/IDEADEV/Demetra+Roadmap

------------------
Maxim Shafirov
JetBrains, Inc
http://www.jetbrains.com
"Develop with pleasure!"

Oh yeah, an option to run Ant targets by default in the background
instead of having to always hit the "background" button.
--
-

David H. McCoy


-

In article <488e3f9b115438c7b9074a7ca250@news.jetbrains.com>,
max@jetbrains.com says...

Hello Amnon,

What about extending support for JavaScript?

It's there on the last line of the map.

What about integrating 'going-back' technology

Would you please explain in a bit more details. I'm affraid I do not follow.

Could be something Visual C++ 6 had. If I'm on line 50, you could right
click, go to line 45, change a value step back to 50 with the change in
place.

And default background execution of Ant targets! :)

it will add complexity to managing IntelliJ

We sincerely hope for exactly opposite!

------------------
Maxim Shafirov
JetBrains, Inc
http://www.jetbrains.com
"Develop with pleasure!"

--
-

David H. McCoy


-

Hello David,

DM> Oh yeah, an option to run Ant targets by default in the background
DM> instead of having to always hit the "background" button.

That's been there from the beginning, check the properties dialog for your
ant script.

HTH,
Andrei

I see your point ... but to not even mention AOP in the roadmap is a
clear signal that AOP is not important for IDEA. I think this is a
chicken/egg problem:

Currently IDEA doesn't support AOP, so IDEA users don't use AOP. Of
course those who need AOP have already switched to Eclipse a long time
ago. Now if Jetbrains build their roadmap for IDEA 6.0 based on what
current IDEA users fancy, it's obvious that only few people will mention
AOP support. But that is no indication of how attractive a feature AOP
support would be. You'd have to ask Eclipse users as well: "Which
features keep you from switching to IDEA".

The bottom line is: If you're thinking about using AOP, don't use IDEA.
I understand that now, and I won't keep asking for it in this newsgroup.
It just means that I'll have to look for a more modern IDE, that's all. ;)

Michael Riess wrote on 17/11/05 23:57:

>> so you think that Aspects are not important? IMO they're the single
>> most important new technological concept since object oriented
>> programming.

AOP is important, supporting AspectJ -ONLY- at the core is not.

That's been there from the beginning, check the
properties dialog for your
ant script.

Key words here are by default. Everytime I add a build script I have to go to the properties dialog and check the "build in background" option. Frankly, I can't see a reason why this isn't on by default. Does anyone here enjoy the progress dialog that much?

I think I first asked for default ant configuration settings about, let's see, four years ago. It's not just the ridiculous choice for background default, it's also continuously having to add common libs to the classpath.

When the default level of usability polish is so high, little things like this do stand out and wrankle.

--Dave Griffith

Hello Michael,

MR> I see your point ... but to not even mention AOP in the roadmap is a
MR> clear signal that AOP is not important for IDEA. I think this is a
MR> chicken/egg problem:
MR>
MR> Currently IDEA doesn't support AOP, so IDEA users don't use AOP.

Says who? You don't need AOP support in idea to use AOP. I use the AOP
facilities with the Springframework.

R

Thanks Robert,

but Spring is a little oversized for me - I don't use J2EE. I guess I'm
not the typical IDEA user ...

Hello Michael,

MR> I see your point ... but to not even mention AOP in the roadmap is a
MR> clear signal that AOP is not important for IDEA. I think this is a
MR> chicken/egg problem:
MR> MR> Currently IDEA doesn't support AOP, so IDEA users don't use AOP.
Says who? You don't need AOP support in idea to use AOP. I use the AOP
facilities with the Springframework.

R

You'd have to be pretty dimwitted to think that AOP == AspectJ and a special compiler. There are a ton of AOP libraries out there that do not require compile time support, and kick in at runtime. Check out dynaop for one.

Either way, why don't you make good on your threat and switch to eclipse and stop annoying the rest of us? Threatening that constantly but never actually doing it (quietly, like a man, instead of whining and crying like a girl) just makes you look silly.

The one good thing I guess from your attempt at ruining this thread is that it does convince everyone else that AOP guys are pretty crazy and should never be taken seriously. You're so 2003.

Wow, your own personal Bile. Some guys would pay good money for that. You should have it framed.

--Dave Griffith

Marcus Brito wrote:
>>That's been there from the beginning, check the
>>properties dialog for your
>>ant script.

Key words here are by default. Everytime I add a build script I have to go to the properties dialog and check the "build in background" option. Frankly, I can't see a reason why this isn't on by default. Does anyone here enjoy the progress dialog that much?

I agree, I think almost every modal dialog should be a toolwindow, permanently.

David H. McCoy wrote:

Could be something Visual C++ 6 had. If I'm on line 50, you could right
click, go to line 45, change a value step back to 50 with the change in
place.

I'm afraid, it won't be possible unless JB implement their own JVM. The
reason is that in Java you cannot manipulate the instruction pointer
(for native apps it's just another CPU register.)

OTOH, as far as I know it's possible to pop frames, so you can restart
from the beginning of the current method. This can have nasty side
effects if one is not careful, but I can imagine it being useful sometimes.

Dimitar

Ease of migration is one thing, but there need to be other, very good reasons for people to abandon their custom-built solutions. Demonstrable and compelling superiority is required to overcome such inertia.

Ciao,
Gordon

I'd buy a gallon!

Dave Griffith wrote:

Wow, your own personal Bile. Some guys would pay good money for that. You should have it framed.

>

--Dave Griffith

How about improvements to Subversion support / UI? I think the Subversion support was somewhat minimal. Improving the setup / configuration of Subversion and improving the history viewing are two areas that could stand some improvement.

It would be wonderful for
IDEA to be able to treat the code as real code,
allowing navigation, etc.

How about sticky notes? The reviewer could select some code and add a sticky note (like "hey dummy, this code sucks!" :)

How about a refactoring called "explode" to turn:

a().b().c(d()).e();

into:

A a = a();
B b = b();
D d = d();
C c = c(d);
E e = e();

That way, setting a breakpoint at the right spot would be easy.

Of course, we'd need a way to undo it, but perhaps ctrl-J would do it just fine.

Hello Erik,

What logic should be followed in assigning variable names?

How about a refactoring called "explode" to turn:

a().b().c(d()).e();

into:

A a = a();
B b = b();
D d = d();
C c = c(d);
E e = e();
That way, setting a breakpoint at the right spot would be easy.

Of course, we'd need a way to undo it, but perhaps ctrl-J would do it
just fine.

Same as "Introduce variable" default, maybe?

What logic should be followed in assigning variable names?

>> How about a refactoring called "explode" to turn:
>>
>> a().b().c(d()).e();
>>
>> into:
>>
>> A a = a();
>> B b = b();
>> D d = d();
>> C c = c(d);
>> E e = e();
>> That way, setting a breakpoint at the right spot would be easy.
>> Of course, we'd need a way to undo it, but perhaps ctrl-J would do it
>> just fine.
>>

Keith Lea <keith@cs.oswego.edu> wrote:

I think it will be very important for many Eclipse users, for JB to
provide hosting for these servers. It's hard to find a web host that even
has Java, let alone a host that would let you set up your own server.

A server with root access is less than $50/month. Even less if it only
is a virtual server. That should be within reach for a team (if the
server is not set up on the internal network). You even get a free
battery powered drilling machine from your web hoster with it...

Dirk Dittert

Hello Maxim,

What about Resharper IDE-style JavaDoc?
http://www.jetbrains.net/confluence/download/attachments/13060/resharper20.jpg

Same as "Introduce variable" default, maybe?

I think that would work. You can always rename them later. And you'll probably want de-explode them when you're done debugging anyway.

It should be noted that there already inspections for chained and nested method calls, with "Introduce Variable" as a quickfix.

--Dave Griffith

It should be noted that there already inspections for chained and
nested method calls, with "Introduce Variable" as a quickfix.

If there was a simple way to use this "automatic unrolling" by other inspections,
it would go great with my #1 suggested inspection:
Eliminate similar tails from if-else statments. E.g. being able to change

if (condition) {
doStuff();
foo(bar(1));
} else {
doOtherStuff();
foo(bar(2));
}

to

int x;
if (condition) {
doStuff();
x = 1;
} else {
doOtherStuff();
x=2;
}
foo(bar(x));

--Dave Griffith

Hello Taras,

This will require re-implementing editor component since it is limited to
editing plain text right now so it doesn't worth efforts IMO.

-

Maxim Shafirov
JetBrains, Inc
http://www.jetbrains.com
"Develop with pleasure!"

Hello Maxim,

What about Resharper IDE-style JavaDoc?
http://www.jetbrains.net/confluence/download/attachments/13060/resharp
er20.jpg