Maybe this is widespread knowledge but Alexander Stepanov (creator of the STL) has some lecture notes online about his take on programming and such, including of course a healthy dose of the thinking that went into creating the STL.

http://www.stepanovpapers.com/notes.pdf

Was mentioned on another list I'm on.  They look interesting at first glance, but I haven't had a chance to look too closely.  Might be especially interesting reading for those currently designing standard libraries. ;-)

--bb

Bill Baxter wrote:
> Maybe this is widespread knowledge but Alexander Stepanov (creator of the STL) has some lecture notes online about his take on programming and such, including of course a healthy dose of the thinking that went into creating the STL.
> 
> http://www.stepanovpapers.com/notes.pdf
> 
> Was mentioned on another list I'm on.  They look interesting at first glance, but I haven't had a chance to look too closely.  Might be especially interesting reading for those currently designing standard libraries. ;-)

Er, just to make sure that's not taken the wrong way -- by ";-)" I do *not* mean to suggest that the folks writing Tango are in need of any design advice.  I just meant that despite referring vaguely to "people working on standard libraries", I really meant specifically Tango folks, since everybody knows there's really only one big standard library project going on at the moment, and it is Tango.  It was a lame attempt at humor, at best.  Hope nobody took it the wrong way.

--bb

Bill Baxter wrote:
> Maybe this is widespread knowledge but Alexander Stepanov (creator of the STL) has some lecture notes online about his take on programming and such, including of course a healthy dose of the thinking that went into creating the STL.

Ah nice, I don't think I've read this one before.  Alexander is an eloquent writer, I suspect it will be a good read :-)


Sean

On Fri, 09 Feb 2007 01:35:31 +0900, Bill Baxter wrote:

> Bill Baxter wrote:
>> Maybe this is widespread knowledge but Alexander Stepanov (creator of the STL) has some lecture notes online about his take on programming and such, including of course a healthy dose of the thinking that went into creating the STL.
>> 
>> http://www.stepanovpapers.com/notes.pdf
>> 
>> Was mentioned on another list I'm on.  They look interesting at first glance, but I haven't had a chance to look too closely.  Might be especially interesting reading for those currently designing standard libraries. ;-)
> 
> Er, just to make sure that's not taken the wrong way -- by ";-)" I do *not* mean to suggest that the folks writing Tango are in need of any design advice.  I just meant that despite referring vaguely to "people working on standard libraries", I really meant specifically Tango folks, since everybody knows there's really only one big standard library project going on at the moment, and it is Tango.  It was a lame attempt at humor, at best.  Hope nobody took it the wrong way.
> 
> --bb


I just wanted to clear something up here concerning the use of the term "standard":

Tango is certainly intended to be "standard" capable, but by no means has
any claims to being a standard unless someone like Walter makes it so
or people implicitly adopt it as such (in this case, that's still not
/really/ standard other than in use). It's really counterproductive to
pretend a library is going to be a standard just because it exhists: I
think there can only be one standard for D (or else "standard" really
becomes a meaningless term if there are multiples of them) and that spot
isn't something with which to trifle. So, Tango is merely an
excellent alternative. I'm biased to think it rather as the better of the
two for various reasons. :) But having a choice is good for people.  And I
think that that is precisely what Tango is trying to express: people
should be given "choice". :)

Tango definitely tries to be what Phobos can not or will not be.
Tango is designed to be community driven -- hightly responsive to bug
reports, suggestions, critiques, and contributions; well documented, well
maintained.  The development model is different than phobos' and really
not comparable; the designs also are noticeably different, Tango's being
noticeably more object-oriented than procedural.  In addtion, Tango is
/painstakingly/ designed for reliability, performance, and small binary
size (I've seen the developers at work at this process and the amount of
energy they put into finding solutions for these issues is
absolutely incredible).

Naturally people have different preferences, and that's fine. Still, I believe Tango can progress much further than Phobos based on the nature of it's development and design.  I also think it will be an excellent library on which to build world-class applications, clients and servers, etc. and to garner a whole lot of attention for the D language, in general. So, it's nothing but good news for D whatever the outcome.

-JJR

(DISCLAIMER: the opinions expressed above indicate my own personal opinion and do not necessarily indicate any opinion of the Tango Team members.  )

John Reimer wrote:
> I
> think there can only be one standard for D (or else "standard" really
> becomes a meaningless term if there are multiples of them) 

But the nice thing about standards is that there are so many of them to choose from... :P

John Reimer wrote:
> I just wanted to clear something up here concerning the use of the term "standard":
> 
> Tango is certainly intended to be "standard" capable, but by no means has
> any claims to being a standard unless someone like Walter makes it so
> or people implicitly adopt it as such (in this case, that's still not
> /really/ standard other than in use). It's really counterproductive to
> pretend a library is going to be a standard just because it exhists: I
> think there can only be one standard for D (or else "standard" really
> becomes a meaningless term if there are multiples of them) and that spot
> isn't something with which to trifle. So, Tango is merely an
> excellent alternative. I'm biased to think it rather as the better of the
> two for various reasons. :) But having a choice is good for people.  And I
> think that that is precisely what Tango is trying to express: people
> should be given "choice". :)

Tango is still different from other libraries in the sense that it is
incompatible with Phobos, and as such is 'a standard' library if that makes
sense, an alternative as you said.

> Tango definitely tries to be what Phobos can not or will not be.
> Tango is designed to be community driven -- hightly responsive to bug
> reports, suggestions, critiques, and contributions; well documented, well
> maintained.  The development model is different than phobos' and really
> not comparable; the designs also are noticeably different, Tango's being
> noticeably more object-oriented than procedural.  In addtion, Tango is
> /painstakingly/ designed for reliability, performance, and small binary
> size (I've seen the developers at work at this process and the amount of
> energy they put into finding solutions for these issues is
> absolutely incredible).

I agree, from what I've seen it is very impressive work, with an attention to detail. You say choice is good, but one thing I'm worried about is that there will be two 'standard libraries', and what this will mean. Some libraries will support Tango, some Phobos, some both. I'm not sure this is a situation that will be good for D in the long run? (For now it might be fine though, it gives Tango a chance to prove itself).

Personally I think Tango has a much better chance and imho it would be best if this is adopted as *the* standard library eventually, if only for the reason there are so many programmers involved.