What are the plans for the Digital Mars D compilers on Windows and Linux? I see in the FAQ that the front ends are open source, and I saw somewhere that the full compilers are "reference implementations" for all D compilers, and I see that they are freely downloadable now, but I still don't understand what this adds up to.

Clearly the D front end plus the GCC compiler will give us a free and open source D compiler for *nix systems, but what about Windows? Will the full DMD compiler remain free of charge for downloading, but not be open source (behind the front end)? Will it still be free of charge but not open source? Or will the Digital Mars compilers cease to be available as free downloads as soon as Version 1.0 officially ships? And on Linux, will DMD and D/GCC both be free of charge but only the latter open source, or what?

Please don't interpret this as a complaint. I'm just trying to understand what (complete) compilers will be available for real production use on which platforms under what terms.

And knowing when would be nice, too. Again, this isn't a complaint. Some projects add a "road map" with rough date estimates to their FAQs. I don't know how much of a road map I would be able to piece together if I were to go back and read the last several thousand newsgroup postings, but it would certainly be more convenient if even a summary of what has already been said were available in a road map of sorts.

Glen Perkins wrote:
> (...)Clearly the D front end plus the GCC compiler will give us a free and open source D compiler for *nix systems, but what about Windows?(...)

I think MinGW + D will make Windows developers happy ;)

Glen Perkins wrote:
> What are the plans for the Digital Mars D compilers on Windows and Linux? I see in the FAQ that the front ends are open source, and I saw somewhere that the full compilers are "reference implementations" for all D compilers, and I see that they are freely downloadable now, but I still don't understand what this adds up to.

Are you aware of the David Friedman's GDC project?
http://home.earthlink.net/~dvdfrdmn/d/

It's based on the open-source front end.


> Clearly the D front end plus the GCC compiler will give us a free and open source D compiler for *nix systems, but what about Windows? Will the full DMD compiler remain free of charge for downloading, but not be open source (behind the front end)? Will it still be free of charge but not open source? Or will the Digital Mars compilers cease to be available as free downloads as soon as Version 1.0 officially ships? And on Linux, will DMD and D/GCC both be free of charge but only the latter open source, or what?

It's my understanding that Walter plans on keeping DMD free. To make money, he plans to sell books and additional tools. But I think the basic DMD compiler will always remain free. But I might not be remembering it right, so maybe we should just wait for Walter to answer.

-- 
Justin (a/k/a jcc7)
http://jcc_7.tripod.com/d/

"J C Calvarese" <jcc7@cox.net> wrote in message news:cn9bii$q6p$1@digitaldaemon.com...
> Glen Perkins wrote:
>> What are the plans for the Digital Mars D compilers on Windows and Linux? I see in the FAQ that the front ends are open source, and I saw somewhere that the full compilers are "reference implementations" for all D compilers, and I see that they are freely downloadable now, but I still don't understand what this adds up to.
>
> Are you aware of the David Friedman's GDC project?
> http://home.earthlink.net/~dvdfrdmn/d/
>
> It's based on the open-source front end.

Yes, that's what I was referring to as covering *nix systems. Of course, in a sense GCC runs on Windows, too. I always have cygwin installed on Windows and GCC installed on any *nix, including cygwin. A compiler in cygwin isn't what I consider a real Windows compiler, though.

Of course, there may also be a fully native Windows GCC that I've forgotten about, since I only use GCC on *nix and use MSVC on Windows.

>> Clearly the D front end plus the GCC compiler will give us a free and open source D compiler for *nix systems, but what about Windows? Will the full DMD compiler remain free of charge for downloading, but not be open source (behind the front end)? Will it still be free of charge but not open source? Or will the Digital Mars compilers cease to be available as free downloads as soon as Version 1.0 officially ships? And on Linux, will DMD and D/GCC both be free of charge but only the latter open source, or what?
>
> It's my understanding that Walter plans on keeping DMD free. To make money, he plans to sell books and additional tools. But I think the basic DMD compiler will always remain free. But I might not be remembering it right, so maybe we should just wait for Walter to answer.

Good plan, but I appreciate your info.

"h3r3tic" <foo@bar.baz> wrote in message news:cn8ta4$8nb$2@digitaldaemon.com...
> Glen Perkins wrote:
>> (...)Clearly the D front end plus the GCC compiler will give us a free and open source D compiler for *nix systems, but what about Windows?(...)
>
> I think MinGW + D will make Windows developers happy ;)

I'm not so sure about that. My impression is that MinGW is for getting *nix apps to run on Windows. (Please correct me if I'm wrong). I've worked at several commercial software companies with well-known shrinkwrapped products and have had dealings with many others, and I don't recall anyone seriously considering using MinGW to build a commercial Windows product. I'm not sure they would be very happy building those apps using a compiler that used minGW, since it would be clear that it wasn't really intended to be a Windows dev tool, and you can't fool around with something as fundamental as your compiler when building a commercial app.

What I *have* seen were cases where something useful from the *nix world was needed for *in-house* use on Windows, and minGW or cygwin were considered viable solutions. (I know I could hardly function on Windows without my cygwin!) The compiler you build your commercial apps with had better be a serious native product, though.

> I'm not so sure about that. My impression is that MinGW is for getting *nix apps to run on Windows. (Please correct me if I'm wrong).

Cygwin was intended for that, yes, but not MingW. It doesn't even have any Posix libraries. It's a port of gcc to Windows for the development of Windows applications, just as DJGPP was intended to build DOS apps. Coding to one tool-chain certainly eases cross-platform development. Though I don't see the benefit of a MingW backend for D, other than to make OSS zealots happy.

Glen Perkins wrote:
> What I *have* seen were cases where something useful from the *nix world was needed for *in-house* use on Windows, and minGW or cygwin were considered viable solutions. (I know I could hardly function on Windows without my cygwin!) The compiler you build your commercial apps with had better be a serious native product, though.

Well, I don't have that much experience with commercial application development (still at uni) but I've worked both with the serious native products (MSVC v.6, 2001,2003) and MinGW. Seriously, I have experienced problems with MSVC6 that were absent in MinGW, and I don't mean standard-inconsistency. This leads me into thinking that MinGW is a well established product that may be used for commercial development. The main factor that I see behind using MSVC is that IMO it's the best IDE ever created. It has really great library support and a huge base of 'faithful' users. But as a compiler I don't see MinGW being worse than Microsoft compilers. In the case of MSVC6 that's still commonly used, I believe MinGW is a much better choice for a COMPILER, maybe its compile times are longer but the resulting code can be faster (I've benchmarked 30% - 100% faster execution of some specific programs of mine).

My conclusion is that D needs a really good IDE that people would work with as this is the main factor for which one vendor is chosen and another rejected. As for the IDE, I obviously don't only mean the code editor, but a RAD environment.

As for writing a MinGW frontend mentioned in another post: Correct me if I'm wrong, but I've always thought that MinGW is 'frontend-compatible' with GCC so the GDC should work with MinGW's backend (maybe with some modifications).

In article <cna0im$1v35$1@digitaldaemon.com>, Mike Parker says...
>
>
>> I'm not so sure about that. My impression is that MinGW is for getting *nix apps to run on Windows. (Please correct me if I'm wrong).
>
>Cygwin was intended for that, yes, but not MingW. It doesn't even have any Posix libraries. It's a port of gcc to Windows for the development of Windows applications, just as DJGPP was intended to build DOS apps. Coding to one tool-chain certainly eases cross-platform development. Though I don't see the benefit of a MingW backend for D, other than to make OSS zealots happy.

Exactly! Through-out this thread, there has been talk of 'free' D compilers. OSS is fine, but free (as in beer) I have a problem with..

I personally think Walter /should/ charge for DMD v1.0, if that means that we'll get a better compiler for it; both because of the commercial incentive that provides as well as the funding it would provide (to more quickly improve the compiler). Alot of the major work on GCC is done through commercial support because then the large sponsors can get their platforms/features supported and still use an unlimited # of 'free' licences for their developers. D doesn't (yet) have that luxury.

It's funny - compilers are some of the most important and complicated software out there, taking arguably the most skill, research, time and resources to produce per LOC, yet many people expect them to be free.

I think a grand marketing plan would be to offer some sort of feature limited (e.g.: hobbled library) trial edition through one download, easily installable on both Windows and *nix, and then charge a nominal fee for the 'full' cross-platform compiler + 'free' minor version upgrades and include all of the tools with the paid-for download.

Assuming that DMD v1.0 will perform as well as GCC C/++, and perform as specified on both Linux and Windows, I would be money (way) ahead by paying $100.00 for a D license rather than debugging some obscure memory leak or segfault because I forgot to NULL a C++ pointer, or because of some obscure incompatibility in the D/GCC interface or MinGW/Cygwin. Heck, I'd be money ahead by buying several licenses rather than dealing with the vagaries of installing Cygwin in support of some portable application (it is very easy to burn many man-hours building, maintaining and supporting complicated installation routines, after all, not to mention the potential for hard/impossible to fix portability issues).

- Dave

In article <cnaok2$3c7$1@digitaldaemon.com>, Dave says...
>
>
>Exactly! Through-out this thread, there has been talk of 'free' D compilers. OSS is fine, but free (as in beer) I have a problem with..

I don't like free software at all. Actually there is no such thing:
DUI is not free, *I* pay for it, *I* pay it for all users,
my family pays for it.
But I don't see an alternative. Even with better equivalent free software
people still uses MS products. Until that monopoly is break I don't
see an alternative to free software.

>It's funny - compilers are some of the most important and complicated software out there, taking arguably the most skill, research, time and resources to produce per LOC, yet many people expect them to be free.

My application is more complicated then your's
(and of course, size - of the application - doesn't matter).

Ant

"J C Calvarese" <jcc7@cox.net> wrote in message news:cn9bii$q6p$1@digitaldaemon.com...
> It's my understanding that Walter plans on keeping DMD free. To make money, he plans to sell books and additional tools. But I think the basic DMD compiler will always remain free. But I might not be remembering it right, so maybe we should just wait for Walter to answer.

That's the plan for the foreseeable future. The DMD source is also licensable for companies that want to create their own commercial D compiler products. Those who want to make GPL use of it can do so for free.