[clean-list] Wish list, part 3

[Rinus Plasmeijer]

Hi Matt,


> > - the IO libraries (and the runtime in general). Maintaining these
> > for different platforms is a lot of work, and not research per se. I
> > understand if the Clean research team does not put an equal amount of
> > work in all versions. Delays in IO library development however,
> > should not hold back new releases of the compiler, and I do think
> > that they should at least put SOME effort in all variants (Linux
> > too). It would be best if additional parties get involved.
>
> My contributions should perhaps be more of a footnote than a note, since I
> am still not completely fluent in Clean, and have little experience with
> developing software or with studying computer science.  Nonetheless,
> regarding IO library support:
>
>     (a)    Is the development of the clean IO library still of sufficient
> theoretical interest to anyone on the research team to merit
> multiple-platform development?

There are some bugs to fix and some improvements to make.
Form a research point of view there are still some interesting questions
with respect to IO, eg
- can we use generic programming techniques for making better IO libraries?
- can we make use of dynamics for obtaining a type safe visual editor?
Besides questions like we already know how to make a multi platform GUI
library,
so there is not a scientific drive for us to make yet another port.


>     (b)    Are the apparent (to my untrained eye) similarities between
> existential data types and 'objects' real and sufficient to allow seamless
> communication between a back-end 'event/message stream interface' written
in
> Clean and a front end 'user interface' written in an object oriented
> programming language?

There are many similarities but also differences...
>
>  If the answers to these question are 'no' and 'yes', respectively, then I
> think the Clean team should:
>     (i)    Declare and finalize a basic GUI standard for multiple-platform
> IO libraries that would
>                 - allow reasonable ease-of-use for experimental and
> prototype programs
>                 - maintain the easy transfer of clean programs at the
> prototype level between platforms

For us the current GUI is more or less fixed (modulo bug fixes).
We have a version for Windows and Mac offering the same interface.
So porting should be very easy.
We do not have resources for making a Linux port (see my previous message in
reply to Marco questions).

>     (ii)    Concentrate on improving the C to Clean interface

If we have time left...

>     (iii)    Redeploy HILT resources to the support of IO libraries for
> platforms that clients need help with. (More profit for HILT -> more
> resources for Clean.)

Due to the fact that I had to amange the department for many years, I was
not able at the same time to put effort in Hilt.
Hilt is willing to do any programming in and for Clean, but we can only do
it on a commercial base.
On request by clients that have to pay for the work.

>     (iv)    Encourage 'open-source' type development of the GUI
> specification and be willing to re-distribute specification consistent and
> platform-transparent libraries(i.e., libraries for which Linux, Mac, and
PC
> versions have all been written) as part of an updated GUI
> specification/library package.

We are willing to do that.

>
> I think that beyond the practical necessity and theoretical interest of
> allowing Clean to interact robustly and provably with that crazy,
> destructively updated external world out there, the development of a
> 'complete' language specific and multi-platform I/O library is a bit of a
> red hampster.  What we have (or will have soon) is sufficient for
> experimentaton and the development of prototype programs.  What we could
> have with a maximum of effort will always lag behind OS-supported
languages
> and visually-oriented design tools for the development the user/system
> interface of 'final' programs.

I agree.

Greetings,

Rinus