digitalmars.D - containers/collections for phobos
- Jonathan M Davis (26/26) Dec 17 2009 I was just wondering what the current plans were for containers/collecti...
- Andrei Alexandrescu (7/9) Dec 18 2009 After much discussions in this group and a lot of reflection, I believe
- Jonathan M Davis (4/15) Dec 18 2009 That's great to hear. I look forward to it. D just keeps getting better ...
- Steven Schveighoffer (7/16) Dec 18 2009 I'm not sure if it will be acceptable for inclusion into Phobos, but
- Tim Matthews (5/21) Dec 18 2009 I haven't actually looked at that code but my thoughts on this anyway:
- KennyTM~ (2/18) Dec 19 2009 At least you need to use the Boost license. ;)
I was just wondering what the current plans were for containers/collections in phobos. Right now, there aren't any. It is, in my opinion at least, by far the largest drawback to phobos. Certainly, I think that that's the greatest advantage that tango has over phobos. Container classes are one of the first things that I look for in a standard library, and in that respect, phobos is sorely lacking. Phobos has some great stuff in it. It even has highly innovative stuff like ranges which would be fantastic with containers/collections, but it lacks the container classes themselves. Granted, D arrays are absolutely fantastic (by far the best arrays that I've ever seen), but even with them being resizable and being able to be used as hash tables, they're no replacement for vectors, linked list, sorted maps, sets (sorted or otherwise), etc. Are there any plans to add container/collection classes to phobos? I assume that they haven't been written yet because there have been higher priorities, but especially if we're looking to have D2 completed sometime in the near future, I would think that we'd want containers to be part of the standard library. Certainly, it's a major caveat for anyone looking to use D. At the moment, you either have to use 3rd party libraries or roll your own, and that's not exactly the ideal situation for something as basic and essential as container classes. - Jonathan M Davis P.S. I must say though, that ranges greatly simplify the writing of container classes. Most of the difficulties that I've had in writing my own have been in putting together good iterator implementations and everything that goes with that. Ranges could really set D containers/collections apart. They're so much like iterators, and yet so much better.
Dec 17 2009
Jonathan M Davis wrote:I was just wondering what the current plans were for containers/collections in phobos.After much discussions in this group and a lot of reflection, I believe I know exactly what there is to be done. It's now a matter of finding time to sit down and write the code. Most definitely there will be details that I hadn't thought of that will jitter the design quite a bit, but I don't expect anything dramatic. Andrei
Dec 18 2009
Andrei Alexandrescu wrote:Jonathan M Davis wrote:That's great to hear. I look forward to it. D just keeps getting better and better. - Jonathan M DavisI was just wondering what the current plans were for containers/collections in phobos.After much discussions in this group and a lot of reflection, I believe I know exactly what there is to be done. It's now a matter of finding time to sit down and write the code. Most definitely there will be details that I hadn't thought of that will jitter the design quite a bit, but I don't expect anything dramatic. Andrei
Dec 18 2009
On Fri, 18 Dec 2009 00:01:40 -0500, Jonathan M Davis <jmdavisProg gmail.com> wrote:I was just wondering what the current plans were for containers/collections in phobos. Right now, there aren't any. It is, in my opinion at least, by far the largest drawback to phobos. Certainly, I think that that's the greatest advantage that tango has over phobos. Container classes are one of the first things that I look for in a standard library, and in that respect, phobos is sorely lacking.I'm not sure if it will be acceptable for inclusion into Phobos, but dcollections *will* be ported to D2. It already is a viable option for D1 phobos. http://www.dsource.org/projects/dcollections -Steve
Dec 18 2009
On 19/12/2009 3:12 p.m., Steven Schveighoffer wrote:On Fri, 18 Dec 2009 00:01:40 -0500, Jonathan M Davis <jmdavisProg gmail.com> wrote:I haven't actually looked at that code but my thoughts on this anyway: If ranges are to stay then if possible I would like to see any collection implementation designed to work well along side them rather than many different inconsistent ways to achieve the same thing.I was just wondering what the current plans were for containers/collections in phobos. Right now, there aren't any. It is, in my opinion at least, by far the largest drawback to phobos. Certainly, I think that that's the greatest advantage that tango has over phobos. Container classes are one of the first things that I look for in a standard library, and in that respect, phobos is sorely lacking.I'm not sure if it will be acceptable for inclusion into Phobos, but dcollections *will* be ported to D2. It already is a viable option for D1 phobos. http://www.dsource.org/projects/dcollections -Steve
Dec 18 2009
On Dec 19, 09 10:12, Steven Schveighoffer wrote:On Fri, 18 Dec 2009 00:01:40 -0500, Jonathan M Davis <jmdavisProg gmail.com> wrote:At least you need to use the Boost license. ;)I was just wondering what the current plans were for containers/collections in phobos. Right now, there aren't any. It is, in my opinion at least, by far the largest drawback to phobos. Certainly, I think that that's the greatest advantage that tango has over phobos. Container classes are one of the first things that I look for in a standard library, and in that respect, phobos is sorely lacking.I'm not sure if it will be acceptable for inclusion into Phobos, but dcollections *will* be ported to D2. It already is a viable option for D1 phobos. http://www.dsource.org/projects/dcollections -Steve
Dec 19 2009