From: Nicolas Cannasse <warplayer@fr...>  20030226 02:01:49

Hi extlist, Here's my mutable list interface. Implementation is available and mostly use the standard List library function. Mutable lists with in place modifications are very usable each time you were previously using "list ref" I'm not putting any documentation with the interface, since every not obvious function can be considered as to be removed or changed. It also provide a set of functions based on item index in the list ( i'm using this, but i'm not really sure it should be in the ExtLib MList... or perhaps in an inner module ) I'm waiting for your comments. Nicolas Cannasse  exception Invalid_index of int exception No_such_element module MList : sig type 'a t val empty : unit > 'a t val isempty : 'a t > bool val copy : 'a t > 'a t > unit val copy_list : 'a t > 'a list > unit val from_list : 'a list > 'a t val to_list : 'a t > 'a list val add : 'a t > 'a > unit val push : 'a t > 'a > unit val pop : 'a t > 'a (* raise No_such_element *) val last : 'a t > 'a (* raise No_such_element *) val first : 'a t > 'a (* raise No_such_element *) val npop : 'a t > int > 'a list (* raise No_such_element *) val clear : 'a t > unit val length : 'a t > int val add_sort : ('a > 'a > int) > 'a t > 'a > unit val hd : 'a t > 'a (* raise No_such_element *) val tl : 'a t > 'a list (* raise No_such_element *) val iter : ('a > unit) > 'a t > unit val find : ('a > bool) > 'a t > 'a val find_ex : ('a > bool) > 'a t > exn > 'a val exists : ('a > bool) > 'a t > bool val map : ('a > 'b) > 'a t > 'b list val sort : ('a > 'a > int) > 'a t > unit val filter : ('a > bool) > 'a t > unit val shuffle : 'a t > unit val remove : 'a t > 'a > unit (* only one element removed, tested with ( = ), raise Not_found *) val remove_if : ('a > bool) > 'a t > unit (* indexes functions *) val index_of : 'a t > 'a > int (* raise Not_found *) val at_index : 'a t > int > 'a (* raise Invalid_index *) val index_of_with : ('a > bool) > 'a t > int (* raise Not_found *) val set : 'a t > int > 'a > unit (* raise Invalid_index *) val remove_at_index : 'a t > int > unit (* raise Invalid_index *) end 
From: Michal Moskal <malekith@pl...>  20030226 10:13:32

On Wed, Feb 26, 2003 at 11:01:04AM +0900, Nicolas Cannasse wrote: > exception Invalid_index of int > exception No_such_element Maybe Invalid_argument "..." or Failure "..." would be better for both? IMHO this is more in style of existing ocaml std library. > module MList : > sig > type 'a t > > val empty : unit > 'a t > val isempty : 'a t > bool > > val copy : 'a t > 'a t > unit Ordering of arguments is copy dst src? I thought copy src dst, but... > val copy_list : 'a t > 'a list > unit ...this made me think right. Maybe assign instead of copy would be better? It suggests the right ordering of arguments. And I know each function here takes argument it operates as first argument, so this shouldn't be hard to remember. And one more Q : is copy constant time operation? > val from_list : 'a list > 'a t > val to_list : 'a t > 'a list > > val add : 'a t > 'a > unit At front or at end? I belive both versions should be available (i.e. list should be cyclic), since it doesn't give much overhead and gives more possiblities. If only pop/push are supported, why not use Stack module from std library? > val push : 'a t > 'a > unit > val pop : 'a t > 'a (* raise No_such_element *) > val last : 'a t > 'a (* raise No_such_element *) > val first : 'a t > 'a (* raise No_such_element *) > val npop : 'a t > int > 'a list (* raise No_such_element *) > val clear : 'a t > unit > val length : 'a t > int Is this constant time operation? > val add_sort : ('a > 'a > int) > 'a t > 'a > unit > > val hd : 'a t > 'a (* raise No_such_element *) > val tl : 'a t > 'a list (* raise No_such_element *) If other internal representation then 'a list is used this will have to be O(n) operation, so I doubt it's good idea to provide it. > val iter : ('a > unit) > 'a t > unit > val find : ('a > bool) > 'a t > 'a > val find_ex : ('a > bool) > 'a t > exn > 'a > val exists : ('a > bool) > 'a t > bool > val map : ('a > 'b) > 'a t > 'b list val map : ('a > 'b) > 'a t > 'b t val map_list : ('a > 'b) > 'a t > 'b list ? > val sort : ('a > 'a > int) > 'a t > unit > val filter : ('a > bool) > 'a t > unit > > val shuffle : 'a t > unit Hmm? What's that? > > val remove : 'a t > 'a > unit (* only one element removed, tested with > ( = ), raise Not_found *) > val remove_if : ('a > bool) > 'a t > unit > > (* indexes functions *) > > val index_of : 'a t > 'a > int (* raise Not_found *) > val at_index : 'a t > int > 'a (* raise Invalid_index *) > val index_of_with : ('a > bool) > 'a t > int (* raise Not_found *) > > val set : 'a t > int > 'a > unit (* raise Invalid_index *) set x n y takes O(n)? > val remove_at_index : 'a t > int > unit (* raise Invalid_index *) Ditto?  : Michal Moskal ::::: malekith/at/pldlinux.org : GCS {C,UL}++++$ a? !tv : PLD Linux ::::::: Wroclaw University, CS Dept : {E,w} {b++,e}>+++ h 
From: Eric C. Cooper <ecc@cm...>  20030226 15:36:05

On Wed, Feb 26, 2003 at 11:09:54AM +0100, Michal Moskal wrote: > Ordering of arguments is copy dst src? I thought copy src dst, but... > > > val copy_list : 'a t > 'a list > unit > > ...this made me think right. > > Maybe assign instead of copy would be better? It suggests the right > ordering of arguments. And I know each function here takes argument it > operates as first argument, so this shouldn't be hard to remember. This is a good case for copy ~src: x ~dst: y (remembering strcpy, bcopy, memcpy, ... in C)  Eric C. Cooper e c c @ c m u . e d u 
From: Brian Hurt <brian.hurt@ql...>  20030228 17:45:42

Hello everyone. Sorry for the silence I had a cold. On Wed, 26 Feb 2003, Michal Moskal wrote: > On Wed, Feb 26, 2003 at 11:01:04AM +0900, Nicolas Cannasse wrote: > > val copy : 'a t > 'a t > unit > > Ordering of arguments is copy dst src? I thought copy src dst, but... This is one of those things which, IMHO, should be standardized all library functions do it the same way. I can see it woring either way: copy src1 src2 src3 ... dst copy dst src1 src2 src3 ... both work. I just want to pick *one* and stick with it. Brian 
From: fva <fva@ts...>  20030226 17:00:55

Nicolas Cannasse wrote: >Hi extlist, > >Here's my mutable list interface. >Implementation is available and mostly use the standard List library >function. > So this is a candidate for a new module for the extlib, *not* an extension for a particular lib in StdLib? >Mutable lists with in place modifications are very usable each time you were >previously using "list ref" > >I'm not putting any documentation with the interface, since every not >obvious function can be considered as to be removed or changed. >It also provide a set of functions based on item index in the list ( i'm >using this, but i'm not really sure it should be in the ExtLib MList... or >perhaps in an inner module ) > >I'm waiting for your comments. > I will try to be constructive and please do not be fooled by typed language... It is very difficult to convey constructive criticism: 1) If your submission was intended as a candidate module, I guess that it should come along with at least documentation for the signature primitives... This is the minimal documentation to show people how to use your module... (I know it's a pain in the neck... I'm only concerned that nobody would take the trouble of using it. Even though it's in the *supermagnificent* :) extlib. 2) If it was intended as an implementation for lists but with mutable characteristics... a) Why not make the connection more evident in the code? b) Why didn't you characterize the complexity of the functions (I guess this is what Moskal did with the questions he put to you) If the answer to both question is "because it was just an example", may I suggest that we *encourage* people submitting either: a) signatures: to state the purpose & conventions of the functions (preferrably in Ocamldoc, but *any* documentation would feel OK with me) b) or implementations: to be articulate as to the running complexities expected of the implementations in either of two ways: a) this function runs in O(whatever)... b) I haven't worked our or tested the complexity... Because I think this would improve *any* code's usability *enormously*. This could be done incrementally and cooperatively with users & such afterwards, but the original author should submit as much information as he is able about the "creature". Regards, Fran Valverde PS: On the technical aspect, yes, I would also try to mimic Stdlib' s way of using exceptions (like very few and overloaded so as to ease their percolation to the main library). F. > >Nicolas Cannasse > > > >exception Invalid_index of int >exception No_such_element > >module MList : > sig > type 'a t > > val empty : unit > 'a t > val isempty : 'a t > bool > > val copy : 'a t > 'a t > unit > val copy_list : 'a t > 'a list > unit > val from_list : 'a list > 'a t > val to_list : 'a t > 'a list > > val add : 'a t > 'a > unit > val push : 'a t > 'a > unit > val pop : 'a t > 'a (* raise No_such_element *) > val last : 'a t > 'a (* raise No_such_element *) > val first : 'a t > 'a (* raise No_such_element *) > val npop : 'a t > int > 'a list (* raise No_such_element *) > val clear : 'a t > unit > val length : 'a t > int > val add_sort : ('a > 'a > int) > 'a t > 'a > unit > > val hd : 'a t > 'a (* raise No_such_element *) > val tl : 'a t > 'a list (* raise No_such_element *) > val iter : ('a > unit) > 'a t > unit > val find : ('a > bool) > 'a t > 'a > val find_ex : ('a > bool) > 'a t > exn > 'a > val exists : ('a > bool) > 'a t > bool > val map : ('a > 'b) > 'a t > 'b list > val sort : ('a > 'a > int) > 'a t > unit > val filter : ('a > bool) > 'a t > unit > > val shuffle : 'a t > unit > > val remove : 'a t > 'a > unit (* only one element removed, tested with >( = ), raise Not_found *) > val remove_if : ('a > bool) > 'a t > unit > >(* indexes functions *) > > val index_of : 'a t > 'a > int (* raise Not_found *) > val at_index : 'a t > int > 'a (* raise Invalid_index *) > val index_of_with : ('a > bool) > 'a t > int (* raise Not_found *) > > val set : 'a t > int > 'a > unit (* raise Invalid_index *) > val remove_at_index : 'a t > int > unit (* raise Invalid_index *) > >end > > 
From: Michal Moskal <malekith@pl...>  20030226 17:16:33

On Wed, Feb 26, 2003 at 06:07:20PM +0100, fva wrote: > 1) If your submission was intended as a candidate module, I guess that > it should come along with at least documentation for the signature > primitives... I guess this was the idea. In fact I figured out sementics of all functions in a minute or two, I just wonder about complexities.  : Michal Moskal ::::: malekith/at/pldlinux.org : GCS {C,UL}++++$ a? !tv : PLD Linux ::::::: Wroclaw University, CS Dept : {E,w} {b++,e}>+++ h 