• teqDruid <me teqdruid.com> Jul 26 2004
• "Ben Hinkle" <bhinkle mathworks.com> Jul 26 2004
• teqDruid <me teqdruid.com> Jul 26 2004
• "Ben Hinkle" <bhinkle mathworks.com> Jul 27 2004
• "Matthew" <admin.hat stlsoft.dot.org> Jul 28 2004
• Gold Dragon <dragonwing dragonu.net> Jul 29 2004
• "Matthew" <admin.hat stlsoft.dot.org> Jul 29 2004

I know it's getting close, but when should we expect DTL?  Are you
allowing people to look at prereleases? If not, could you be?

I ask because I am right now writing code that is in desperate need of
DTL.  I'm using primitive arrays, where I should be using sets, and this
is architectural code (like method signatures that'll be used everywhere),
so I don't need a fully featured or fully working DTL, just basically a
set of prototypes and skeleton code, which I'm sure is done.

I don't want to delay writing this stuff any more than I already have, and
I don't want to rewrite it once DTL is out either.  Something of a pickle.

Thanks
John

"teqDruid" <me teqdruid.com> wrote in message
news:pan.2004.07.26.15.35.22.913434 teqdruid.com...
 I know it's getting close, but when should we expect DTL?  Are you
 allowing people to look at prereleases? If not, could you be?

 I ask because I am right now writing code that is in desperate need of
 DTL.  I'm using primitive arrays, where I should be using sets, and this
 is architectural code (like method signatures that'll be used everywhere),
 so I don't need a fully featured or fully working DTL, just basically a
 set of prototypes and skeleton code, which I'm sure is done.

 I don't want to delay writing this stuff any more than I already have, and
 I don't want to rewrite it once DTL is out either.  Something of a pickle.

 Thanks
 John


I know this doesn't answer your question but mintl.map can be used as a set
if you just pick some random value type - eg Map!(K,int). Or if you don't
need sorting a builtin assoc array int[K] would also do. My gut tells me it
is too early in D's existance to consider any API as "set in stone" - we
don't even have any APIs written on tissue paper much less stone. So I'd
recommend being flexible.

-Ben

On Mon, 26 Jul 2004 15:58:12 -0400, Ben Hinkle wrote:

 
 "teqDruid" <me teqdruid.com> wrote in message
 news:pan.2004.07.26.15.35.22.913434 teqdruid.com...
 I know it's getting close, but when should we expect DTL?  Are you
 allowing people to look at prereleases? If not, could you be?

 I ask because I am right now writing code that is in desperate need of
 DTL.  I'm using primitive arrays, where I should be using sets, and this
 is architectural code (like method signatures that'll be used
 everywhere), so I don't need a fully featured or fully working DTL, just
 basically a set of prototypes and skeleton code, which I'm sure is done.

 I don't want to delay writing this stuff any more than I already have,
 and I don't want to rewrite it once DTL is out either.  Something of a
 pickle.

 Thanks
 John


 I know this doesn't answer your question but mintl.map can be used as a
 set if you just pick some random value type - eg Map!(K,int). Or if you
 don't need sorting a builtin assoc array int[K] would also do. My gut
 tells me it is too early in D's existance to consider any API as "set in
 stone" - we don't even have any APIs written on tissue paper much less
 stone. So I'd recommend being flexible.
 
 -Ben


There are all sorts of things that I *could* do, and I am doing, but it
would be nice to have something that I'll keep using.  I'm assuming that
DTL will become pretty standard, so I'd like to start using it- even if
it's just a skeleton implementation.

Basically, if DTL is interface based (defines interfaces, and implements
them) it'd be nice if the interfaces were released soon (or now).

John

"Ben Hinkle" <bhinkle mathworks.com> wrote in message
news:ce3nol$1ecd$1 digitaldaemon.com...
 "teqDruid" <me teqdruid.com> wrote in message
 news:pan.2004.07.26.15.35.22.913434 teqdruid.com...
 I know it's getting close, but when should we expect DTL?  Are you
 allowing people to look at prereleases? If not, could you be?

 I ask because I am right now writing code that is in desperate need of
 DTL.  I'm using primitive arrays, where I should be using sets, and this
 is architectural code (like method signatures that'll be used




 so I don't need a fully featured or fully working DTL, just basically a
 set of prototypes and skeleton code, which I'm sure is done.

 I don't want to delay writing this stuff any more than I already have,




 I don't want to rewrite it once DTL is out either.  Something of a




 Thanks
 John


 I know this doesn't answer your question but mintl.map can be used as a


 if you just pick some random value type - eg Map!(K,int). Or if you don't
 need sorting a builtin assoc array int[K] would also do. My gut tells me


 is too early in D's existance to consider any API as "set in stone" - we
 don't even have any APIs written on tissue paper much less stone. So I'd
 recommend being flexible.

 -Ben


In fact I've attached a little set.d for making wrappers around Map!(K,int)
or int[K] depending on if sorting is needed or not. The one usability issue
with this approach is that the "int" will show up in foreach statements or
other signatures. But it works pretty well in general. I'll probably expand
out the doc and dump it into mintl.

-Ben

I'm still working away, whenever I can grab the time away.

I've spent the whole day on it today, and I intend to release *something*
before I go to bed tonight, but I tell you
it's incredibly frustrating work because the compiler is still so full of bugs.
Plenty of times I'm having to think of
neat workarounds, or just leaving out swathes of functionality, because either
the compiler crashes, or the exe won't
link, or the exe crashes.

FTR, I've got List, Map, Vector, Stack, and Queue, in varying degrees of
completeness, and may have Set done also soon.
By "varying degrees of completeness", this means whether or not they support
the different enumeration models: foreach,
ranges, Object-interfaces and parameterised interfaces. I'll provide a table of
completeness in the docs when I release,
so I'm not seen to be promising functionality that does not exist.

I apologise _again_ to everyone who's (still) interested, but I promise I'm not
just wagging my tail out here.

Anyway, I'm getting to the point where the frustrations of working with the
current state of the compiler/language are
becoming quite demotivating, so I intend to get out what I can in the next day
or so, and then let all the many smart
minds out there have a look, and hopefully pick up things I've missed.

Cheers

Matthew


"teqDruid" <me teqdruid.com> wrote in message
news:pan.2004.07.26.15.35.22.913434 teqdruid.com...
 I know it's getting close, but when should we expect DTL?  Are you
 allowing people to look at prereleases? If not, could you be?

 I ask because I am right now writing code that is in desperate need of
 DTL.  I'm using primitive arrays, where I should be using sets, and this
 is architectural code (like method signatures that'll be used everywhere),
 so I don't need a fully featured or fully working DTL, just basically a
 set of prototypes and skeleton code, which I'm sure is done.

 I don't want to delay writing this stuff any more than I already have, and
 I don't want to rewrite it once DTL is out either.  Something of a pickle.

 Thanks
 John

Well, I'm far from a smart mind but I'm a smart ass and have a way of 
working around crap or throwing stuff against walls and giving up. 
Either one is cool and perhaps speaking my mind but I guess I should be 
nice since you may or may not release it tonight at a certain level of 
completeness.

 I'm still working away, whenever I can grab the time away.
 
 I've spent the whole day on it today, and I intend to release *something*
before I go to bed tonight, but I tell you
 it's incredibly frustrating work because the compiler is still so full of
bugs. Plenty of times I'm having to think of
 neat workarounds, or just leaving out swathes of functionality, because either
the compiler crashes, or the exe won't
 link, or the exe crashes.
 
 FTR, I've got List, Map, Vector, Stack, and Queue, in varying degrees of
completeness, and may have Set done also soon.
 By "varying degrees of completeness", this means whether or not they support
the different enumeration models: foreach,
 ranges, Object-interfaces and parameterised interfaces. I'll provide a table
of completeness in the docs when I release,
 so I'm not seen to be promising functionality that does not exist.
 
 I apologise _again_ to everyone who's (still) interested, but I promise I'm
not just wagging my tail out here.
 
 Anyway, I'm getting to the point where the frustrations of working with the
current state of the compiler/language are
 becoming quite demotivating, so I intend to get out what I can in the next day
or so, and then let all the many smart
 minds out there have a look, and hopefully pick up things I've missed.
 
 Cheers
 
 Matthew

I'm still drowning under compiler bugs. (See my rant on the digitalmars.D ng)

"Gold Dragon" <dragonwing dragonu.net> wrote in message
news:ceagec$192k$1 digitaldaemon.com...
 Well, I'm far from a smart mind but I'm a smart ass and have a way of
 working around crap or throwing stuff against walls and giving up.
 Either one is cool and perhaps speaking my mind but I guess I should be
 nice since you may or may not release it tonight at a certain level of
 completeness.

 I'm still working away, whenever I can grab the time away.

 I've spent the whole day on it today, and I intend to release *something*
before I go to bed tonight, but I tell you
 it's incredibly frustrating work because the compiler is still so full of
bugs. Plenty of times I'm having to think




 neat workarounds, or just leaving out swathes of functionality, because either
the compiler crashes, or the exe




 link, or the exe crashes.

 FTR, I've got List, Map, Vector, Stack, and Queue, in varying degrees of
completeness, and may have Set done also




 By "varying degrees of completeness", this means whether or not they support
the different enumeration models:




 ranges, Object-interfaces and parameterised interfaces. I'll provide a table
of completeness in the docs when I




 so I'm not seen to be promising functionality that does not exist.

 I apologise _again_ to everyone who's (still) interested, but I promise I'm
not just wagging my tail out here.

 Anyway, I'm getting to the point where the frustrations of working with the
current state of the compiler/language




 becoming quite demotivating, so I intend to get out what I can in the next day
or so, and then let all the many




 minds out there have a look, and hopefully pick up things I've missed.

 Cheers

 Matthew