From: Martin J. <mar...@em...> - 2004-04-27 11:50:22
|
On Tue, 27 Apr 2004, Nicolas Cannasse wrote: > > Is it supposed to replace the INRIA standard library? > > No. We're positioning ourselves as a proposal for extension. That means that > some parts of the ExtLib might be added to the OCaml stdlib if INRIA wish to > do so : we would actually be happy about that. Whatever happens, we'll still > maintain an independant library that can enrich the standard ocaml library > and research additional ways of doing things better. So, what makes you better than them? What additional power do you have? Things like: - hundreds of developers (?) - extra financial support (?) - exceptional communication skills (?) - ... > > Why is there no answer from the SF-ExtLib team when I (or others) post > > some code? This is very disappointing. > > I usually answer most of the messages, all subsribers on this list are also > welcome to do so. If there is no answer to your message it means that maybe > people here are too busy to answer, or maybe that there is no interest on > this subject. > > If you're precisely talking about your recent post "('a,'b list ref) > Hashtbl.t" then I personaly never had the need of such module (although I am > programming ocaml almost everyday in different domains). That does not mean > that it is not useful, but maybe that does mean it's not wide usage enough > to be included. I dismissed myself from answering since I don't see how it > can be useful. If some people were interested by this module, they would > have answered. If there was a common agreement that it's useful (whatever my > opinion) then I would have committed the code into the CVS. There were many messages on the caml-list about a Hashtbl.keys function, and how to remove the redundancies (!). The answer is simple: don't add the same key multiple times. Use Hashtbl.replace if don't want to keep all the previous bindings. If you need to keep them, then use something like my proposal, and everything will become O(1) or O(n). I really think that such a module is useful because it hides an imperative approach behind a safe interface. Sorry, but I thought first that SF-ExtLib was an attempt to provide a more complete library than the standard one. For me, "more complete" does not mean "adding a few functions that INRIA was too stupid to forget", but providing 10 times more modules than INRIA and using the community to maintain this large collection. If only 2 or 3 persons have the power to check and decide what should be in SF-ExtLib, then the result will be similar to the current INRIA's standard library, unless you convince Superman to learn OCaml. ... unless there are very rational and precise conditions for adding things into the library so that the ExtLib leaders can check in 1 minute if some proposal is acceptable or not. I hope you understand my critics and that reading this is not a complete waste of time. Thanks for your attention, Martin |