Re: [Ocaml-lib-devel] Advancements

[Brian H. <bri...@ql...>]

On Mon, 24 Mar 2003, Nicolas Cannasse wrote:

> > - bitset is being rewritten into bitarray, with the new features suggested
> > by the change in name.  The goal is to produce something you could write
> > elliptic curve cryptosystems in at least semi-efficiently.
> 
> Nice !

Status report on this: I spent the weekend writting and rewritting and 
rerewritting the damned rotate functions.  Everything else looks like it's 
about ready to go.  I'm going to take another swing at the rotates 
tonight, and see if I can kick my brain into working.

> 
> > - I have the beginings of what I'm currently calling an XString library,
> > which stores a string as an unbalanced tree of of substrings.  The big
> > advantage here is that concats are O(1), and then converting the XString
> > to a normal string is O(n).  In-place modification won't be supported.
> > This will be usefull in constructing strings by concating substrings
> > efficiently.  Sort of like Java's StringBuffer.
> 
> Very good also, but what then does not call it StringBuffer ?
> The name came by itself, so... :)

I may.

> 
> > - A resizable array module- rather like Java's Vector or Python's List.
> 
> Yes we realy need this Vector one

As a side note, I do way too much numerical programming to like the term
'vector' to denote extensible arrays.  This annoys me no end in Java.  
I'm likely to call this library 'xarray' or something instead.

> 
> > Life has been rather hectic of late for me- sorry I haven't been more
> > timely on this.  But this weekend is looking dull, so I should be able to
> > spend some serious coding time.
> 
> Yes, I would like to apologize for my previous post, and after some
> thinking, I think now we're walking the good way, and we have no need to
> hurry more. 

It's OK.

> I'm currently still working on the List implementation... okay ,
> some words about it : I'm currently at Kyoto, and having everyday at lunch
> some ocaml-talks with Jacques Garrigue, and when I told him about this
> setcdr trick, is first reaction was that it would be nice but perhaps there
> is some GC issues since that can create pointers from the major to the minor
> heap, and then causing speed overhead. 

I'm only using setcdr on list elements I just recently allocated, which 
makes it signifigantly less likely that I would be adding a pointer from 
the major to the minor heap.  Possible, but not likely.

Brian