probability of some events

probability of some events

[Rakotomandimby Mihamina]

Hi,

1°) A bit of history before the question.

Some of my collegues are running Python at the moment.

Python has decided to improve some things, and there is a two step 
improve coming:
  - Python 2.6
  - Python 3

Both of them introduce syntax incompatibility, especially 2.6 on 
exceptions (which is a kind of handicap).

They run Debian Lenny, stuck with Python 2.5 and I run Ubuntu with 
Python 2.6. I should have developped for them.

2°) Now the question

Are you aware of such future changes in OCaml, that would lead to 
incompatibility?

We intend to begin huge developments in OCaml (Web, GUI, system,...)
and such changes will make our task a bit more difficult.

Thank you.

-- 
       Architecte Informatique chez Blueline/Gulfsat:
    Administration Systeme, Recherche & Developpement
                                    +261 34 29 155 34

Re: [Caml-list] probability of some events

[Christophe TROESTLER]

Hi,

On Wed, 9 Sep 2009 08:12:48 +0300, Rakotomandimby Mihamina wrote:
> 
> Are you aware of such future changes in OCaml, that would lead to
> incompatibility?

I of course can't speak for the OCaml core developers but they have
been very cautious not to introduce this kind of incompatibility.

Now, OCaml is also different as it possess the wonderful Camlp4.  In
case such incompatibility would be required to move forward, it is
very likely that a camlp4 syntax extension would be developed to
translate your code automatically as much as possible.

> We intend to begin huge developments in OCaml (Web, GUI, system,...)
> and such changes will make our task a bit more difficult.

Great to hear !

Cheers,
ChriS

RE: [Caml-list] probability of some events

[David Allsopp]

> > Are you aware of such future changes in OCaml, that would lead to
> > incompatibility?

With the usual caveat that past performance is not an indicator of future
wealth...

In the last few years, the only change which caused a bit of an uproar was
camlp4 between 3.09 and 3.10 (which was totally incompatible but had the
same command name). This has all since resolved itself (camlp5 maintained
separately for developers who want the old mechanism and the new camlp4 now
in general use) but I think it's reasonable to say that the response to it
at the time means that it's unlikely that such a breaking change would be
introduced within the 3.x branch of OCaml in the future, but of course I too
don't speak for the guys at Inria. It seemed obvious from the list posts at
the time that 3.10.0 was adopted more slowly than normal minor
version-number releases because of the breaking change in camlp4.

Further history (looking only at Changes in the sources: I started using
OCaml at version 3.06, I think) suggests that breaking changes between major
versions have either been detectable and trivial to fix or aided with
translation tools. For example, stdlib function names had to change between
Caml Special Light 1.15 and Objective Caml 1.00 or new keywords between
OCaml 1.07 and 2.00 and between OCaml 2.04 and 3.00, there appears to have
been much concern for backwards compatibility with tools provided to help
the transition from various different previous versions of the merged
compilers.

As for what gems OCaml 4.x may or may not feature, you'd have to bribe the
priests at Inria for details of their dreams and wishes ;o)

> We intend to begin huge developments in OCaml (Web, GUI, system,...)
> and such changes will make our task a bit more difficult.

Your (valid) worry about future incompatibilities definitely shouldn't put
you off this! It's also painless to run multiple versions of OCaml
side-by-side (at least if building from sources).

HTH,


David

Re: [Caml-list] probability of some events

[Jon Harrop]

On Wednesday 09 September 2009 08:54:08 David Allsopp wrote:
> > > Are you aware of such future changes in OCaml, that would lead to
> > > incompatibility?
>
> With the usual caveat that past performance is not an indicator of future
> wealth...
>
> In the last few years, the only change which caused a bit of an uproar was
> camlp4 between 3.09 and 3.10 (which was totally incompatible but had the
> same command name). This has all since resolved itself (camlp5 maintained
> separately for developers who want the old mechanism and the new camlp4 now
> in general use) but I think it's reasonable to say that the response to it
> at the time means that it's unlikely that such a breaking change would be
> introduced within the 3.x branch of OCaml in the future, but of course I
> too don't speak for the guys at Inria. It seemed obvious from the list
> posts at the time that 3.10.0 was adopted more slowly than normal minor
> version-number releases because of the breaking change in camlp4.

Equality was also changed and caused some problems but I agree that very few 
breaking changes have been made to OCaml in the recent past.

-- 
Dr Jon Harrop, Flying Frog Consultancy Ltd.
http://www.ffconsultancy.com/?e

Re: probability of some events

[Sylvain Le Gall]

Hello,

On 09-09-2009, Rakotomandimby Mihamina <mihamina@gulfsat.mg> wrote:
>
> 2°) Now the question
>
> Are you aware of such future changes in OCaml, that would lead to 
> incompatibility?

OCaml is quite conservative and don't add incompatible features very
easily. I think INRIA team is aware of that and wishing to keep it this
way. 

>
> We intend to begin huge developments in OCaml (Web, GUI, system,...)
> and such changes will make our task a bit more difficult.
>

At least for the 2-3 years coming there will be no problem for sure.
After that, we cannot predict the future ;-) 

But if your concern is stability of language, OCaml seems a good choice.

Regards,
Sylvain Le Gall

Re: [Caml-list] Re: probability of some events

[Rakotomandimby Mihamina]

09/09/2009 11:04 AM, Sylvain Le Gall:
>> We intend to begin huge developments in OCaml (Web, GUI, system,...)
>> and such changes will make our task a bit more difficult.
> At least for the 2-3 years coming there will be no problem for sure.
> After that, we cannot predict the future ;-)

It's enough.
Thanks.

-- 
       Architecte Informatique chez Blueline/Gulfsat:
    Administration Systeme, Recherche & Developpement
                                    +261 34 29 155 34