John Reimer wrote:
> As for standard libraries, I think you should be ready to endorse an organized effort that presents itself with these traits:
>
> 1. Is well documented
> 2. Meets the general approval of the community
> 3. Continues to be developed for multiple compilers in tandem (dmd
> and gdc)
> 4. Is actively developed for multiple platforms (linux, win32, Mac
> OS X)
> 5. Has a strategy layed out for future direction
> 6. Has a dedicated core group of developers that have shown
> dedication to the D language.
> 7. Is maintained under a version control system
>
> Such traits by far surpass what Phobos can offer.  From my perspective, the act of endorsing such an effort hardly constitutes a risk.
>
> -JJR

I wouldn't't limit points 1 and 7 to Phobos. Hosting a publicly readable vcs for DMD's frontent, so that everybody can see individual changes - not only the changes lumped together for each new release - and compile the frontend. This should result in cleaner code* and potentially more contributors.

Thomas

* Extract the attached dmd_stub.zip, fill the dmd directory with the current DMD sources and try to compile it...

Georg Wrede wrote:
> Walter Bright wrote:
>> What I'm mindful of is I endorsed DWT as the official D gui library, which promptly killed it.
> 
> IMHO, more important than having an official GUI library, would be to actually have _something_ right inside dmd.zip!
> 
> It doesn't have to be the world and the kitchen sink, but anybody downloading dmd.zip should be able to write a GUI hello world simply and easily.

At the very least, include the full set of Windows .lib files from the latest SDK, so that at least it's possible to do SDK programming.

John Reimer wrote:
> Shame on you, Walter.  You should know better than to think you
> have that much influence. ;D

If I'm not going to actively manage something, it's best not to get involved with it, as such tends to discourage others from taking the lead on it. They assume I will.


> The momentum died for strictly different reasons (most notable
> being the lack of contributors).

Lack of contributors is why most things wither. The more interesting question is why the lack of contributors. I had (incorrectly) thought that endorsing DWT would lead to more contributors and a more concentrated effort to get it done. The opposite seemed to happen.

> Your choice and timing just
> happened to be off on the matter. Okay, I admit that there was one
> other significant issue that was a personal annoyance: you failed
> to contact or discuss /anything/ with the people considering the
> ports of DWT (myself, Carlos ). You just announced the ports and
> that we were doing them (even though you had no idea what are
> personal feelings on the matter were or how serious we were about
> it).  I recall being quite shocked at your announcement. I think
> Carlos was too.

I'm sorry about that. I had incorrectly just assumed you'd be pleased by it. I wanted to support you guys' efforts.

> Regardless, we all know that GUI Frameworks are particularly
> troublesome to endorse since the area is so subjective.  It's
> probably a lost cause trying to support one over the other.  Best
> to encourage any GUI that people are willing to develop for D
> because I don't think any one framework will be acceptable as a
> standard.

That's where we're at now. There are several D gui's, too many for this community to really support properly.

> As for standard libraries, I think you should be ready to endorse
> an organized effort that presents itself with these traits:
> 
> 1. Is well documented
> 2. Meets the general approval of the community
> 3. Continues to be developed for multiple compilers in tandem (dmd
> and gdc)
> 4. Is actively developed for multiple platforms (linux, win32, Mac
> OS X)
> 5. Has a strategy layed out for future direction
> 6. Has a dedicated core group of developers that have shown
> dedication to the D language.
> 7. Is maintained under a version control system
> 
> Such traits by far surpass what Phobos can offer.  From my
> perspective, the act of endorsing such an effort hardly constitutes
> a risk.

I encourage, and have encouraged, anyone who wants to do this. Any or all parts of Phobos can be used as a starting point. The compiler is my central focus, to enable great libraries to be written.

On Mon, 16 Oct 2006 16:55:55 +1000, Walter Bright <newshound@digitalmars.com> wrote:

>> As for standard libraries, I think you should be ready to endorse
>> an organized effort that presents itself with these traits:
>>  1. Is well documented
>> 2. Meets the general approval of the community
>> 3. Continues to be developed for multiple compilers in tandem (dmd
>> and gdc)
>> 4. Is actively developed for multiple platforms (linux, win32, Mac
>> OS X)
>> 5. Has a strategy layed out for future direction
>> 6. Has a dedicated core group of developers that have shown
>> dedication to the D language.
>> 7. Is maintained under a version control system
>>  Such traits by far surpass what Phobos can offer.  From my
>> perspective, the act of endorsing such an effort hardly constitutes
>> a risk.
>
> I encourage, and have encouraged, anyone who wants to do this. Any or all parts of Phobos can be used as a starting point. The compiler is my central focus, to enable great libraries to be written.

Are you saying that, for example, you would have no problems with Phobos being taken from DigitalMars and moved to DSource so that everyone who needed to could get to submit changes to the library that would actually be applied.

-- 
Derek Parnell

Derek Parnell wrote:
> On Mon, 16 Oct 2006 16:55:55 +1000, Walter Bright 
>> I encourage, and have encouraged, anyone who wants to do this. Any or all parts of Phobos can be used as a starting point. The compiler is my central focus, to enable great libraries to be written.
> 
> Are you saying that, for example, you would have no problems with Phobos being taken from DigitalMars and moved to DSource so that everyone who needed to could get to submit changes to the library that would actually be applied.

The license for the Phobos source code certainly allows that, and that's part of the point of having the license be that way. So no, I have no problem with that.

Georg Wrede wrote:

> IMHO, more important than having an official GUI library, would be to actually have _something_ right inside dmd.zip!
> 
> It doesn't have to be the world and the kitchen sink, but anybody downloading dmd.zip should be able to write a GUI hello world simply and easily.

I think the best bet for *inclusion* with DMD would be MinWin ?
(it has a new home page at http://wiki.dprogramming.com/MinWin)
Tk or wx would work too (it does for Python), but they are much
bigger and I wouldn't want all of Tcl/Tk or wxWidgets included...

The only problem is that it doesn't support Mac OS X, but then
again neither does DMD so I don't think that would matter much ?
It's not a very big issue for Mac users to use the X11* or GTK+
versions of the interface, until a native Mac port can be done.


I am planning to include wxD with my GDC, but I think including
MinWin with DMD would be enough to get started with GUI easily ?

And, very importantly, it does play a game of Classic Empire :-)
(or used to, see http://dsource.org/forums/viewtopic.php?t=660)

--anders

* Always included, see http://www.apple.com/macosx/features/x11/

Derek Parnell wrote:

> Are you saying that, for example, you would have no problems with Phobos  being taken from DigitalMars and moved to DSource so that everyone who  needed to could get to submit changes to the library that would actually  be applied.

I'm confused, wasn't this what the Ares project was all about ?

--anders

Walter Bright wrote:
> Derek Parnell wrote:
>> On Mon, 16 Oct 2006 16:55:55 +1000, Walter Bright
>>> I encourage, and have encouraged, anyone who wants to do this. Any or all parts of Phobos can be used as a starting point. The compiler is my central focus, to enable great libraries to be written.
>>
>> Are you saying that, for example, you would have no problems with Phobos being taken from DigitalMars and moved to DSource so that everyone who needed to could get to submit changes to the library that would actually be applied.
> 
> The license for the Phobos source code certainly allows that, and that's part of the point of having the license be that way. So no, I have no problem with that.

Great to hear you agree.  Before going nuts, can we discuss your feelings towards actually using an externally hosted phobos (or other libraries for that matter) in future dmd release tarballs?  What sort of coordination would you envision?  Would you take select patches and merge with your internal tree?  Abandon your internal tree?  Today you release at your whim.  External development of pieces changes that model.

Assuming a community develops around an externally hosted (be it a fork or the primary) version of phobos, a slightly more mature release process probably ought to develop.  Maybe even introduce the concept of beta versions (which might help with the paper-hat bugs like left in debugging code :)  I'm willing to bet that people like Thomas would be willing to give beta releases a spin.  In his case giving it a whirl through dstress to find regressions before they hit the street.

Later,
Brad

On Mon, 16 Oct 2006 19:02:41 +1000, Anders F Björklund <afb@algonet.se> wrote:

> Derek Parnell wrote:
>
>> Are you saying that, for example, you would have no problems with Phobos  being taken from DigitalMars and moved to DSource so that everyone who  needed to could get to submit changes to the library that would actually  be applied.
>
> I'm confused, wasn't this what the Ares project was all about ?

My understanding is that Ares is a replacement for Phobos. An alternative to Phobos.

Walter seems to be giving the explicit green-light to take the primary responsibility for doing all the coding to Phobos from himself. The way I read it now is that Brad could set up a new DSource project called 'phobos', open it up for read-access for everyone, and svn-commit access to 'registered developers', and a small group to be responsible for creating downloadable library releases. The documentation could also be DSource'd too if Walter doesn't mind submitting the DDOC files he currently uses.

I'm sure this community can come up with a workable collabaration on the 'official' phobos library.

-- 
Derek Parnell

Derek Parnell wrote:

>> I'm confused, wasn't this what the Ares project was all about ?
> 
> My understanding is that Ares is a replacement for Phobos. An alternative  to Phobos.

Yeah, a community alternative to the Phobos library was what I thought.
But I don't really know, and it might have changed, so I won't guess...

I do know that it takes three parts to replace current phobos library:
- the DMD runtime library
- the DMD garbage collector
- the D standard library itself

I guess this would be mostly about Phobos-the-library (std), and a few parts of Deimos (etc) could probably be integrated into Phobos as well ?

> Walter seems to be giving the explicit green-light to take the primary  responsibility for doing all the coding to Phobos from himself. The way I  read it now is that Brad could set up a new DSource project called  'phobos', open it up for read-access for everyone, and svn-commit access  to 'registered developers', and a small group to be responsible for  creating downloadable library releases. The documentation could also be  DSource'd too if Walter doesn't mind submitting the DDOC files he  currently uses.

If we can't even get the GPhobos additions integrated with Phobos, then I'm not so sure that Phobos will be opened up for community development.

I'm hoping that I'm wrong, though. I don't know what this new plan is ?

It *would* be very nice if e.g. the stderr errors and std.c.unix.unix
system module could make it into DMD and Phobos, from GDC and GPhobos.

--anders