Dialects (any plan is better than no plan) - Bane or Blessing?
[1/4] from: tim:johnsons-web at: 27-Sep-2002 11:42
Hello All:
* Gregg Irwin <[greggirwin--mindspring--com]> [020927 08:27]:
> Hi Brett,
> <<...I didn't quite understand what you meant by "In a good dialect, there
<<quoted lines omitted: 8>>
> nomenclature, and any that are well designed, should just kind of "hang
> together" in a very natural way.
Many other programmers have told me that rebol looks very promising,
but shy away because of the lack of a standard library or API. And
frankly I think rebol is "old enough" to have that now. It's too bad
that RT (IMHO) hasn't taken (or delegated) the lead in that.
Lacking that, a series of well-documented dialects might help
to fill that gap.
I've been working with Andrews 'ML dialect, and am gradually phasing
out my own 'html object in favor of 'ML. One of my tendencies in the
beginning of using 'ML was to hack Andrew's base code to add features
that I thought I might want. I revised that approach in favor of
not messing with his code. I would think that if Andrew found any of
my "needs" or "to-dos" of merit, he would then incorporate them.
To make a long story short, if we can't get it together to have a
standard library, it appears the energy is really focused in the 'dialect
area, so let's go for it.
BTW: I wrote an introductory review of 'ML in the September issue of
Frozen North Linux Webzine http://www.frozen-north-linuxonline.com/
and am preparing another review for October. Hey Andrew! stick around
for the weekend, I'm going to have questions for you ... :-)
--
Tim Johnson <[tim--johnsons-web--com]>
http://www.alaska-internet-solutions.com
http://www.johnsons-web.com
[2/4] from: greggirwin:mindspring at: 27-Sep-2002 15:41
Hi Tim, et al
<<
Many other programmers have told me that rebol looks very promising,
but shy away because of the lack of a standard library or API. And
frankly I think rebol is "old enough" to have that now.
>>
I was thinking about this because some time ago I put out a feeler message
to see if people were interested in organizing a set of function libraries.
Not much interest. I have my own, and I know some others do as well, but
there is no standard architecture for them which I think is important. I
started mine when I was very new to REBOL, so I have little faith that my
approach is a very good one, hence I haven't put it out there. In any case,
I hope that when Carl makes an announcment about the new libmaster that
we'll be able to get something going in this area.
As far as the lack of a standard library in REBOL, I have an opinion about
this as well (I'm just full of opinions lately :).
If you compare REBOL to some other languages, there is a lot of comparable
functionality. The difference is that in REBOL they aren't considered
libraries. The various net protocols are a good example. Some languages even
list things as library components that we would think of as integral parts
of the language. If we were to draw a line between native functions versus
schemes, mezzanines, and other pieces written in REBOL, it would probably
look like REBOL had a comparable set of standard libraries. It's just that
we don't see them that way. :)
Maybe what we really need is a document that discusses this issue and
outlines some of the built-in pieces that eliminate the need for many
standard libraries
.
Or maybe a link on www.rebol.com to a page for the "REBOL Standard
Libraries" and a bullet list on that page:
POP3 built-in
HTTP built-in
TCP/Sockets built-in
...
Parsing built-in
Currency built-in
Dates built-in
Serialization built-in
Base64 built-in
Compression built-in
Encryption built-in
GUI built-in (View only, no TK required)
...
<<
It's too bad that RT (IMHO) hasn't taken (or delegated) the lead in that.
Lacking that, a series of well-documented dialects might help
to fill that gap.
>>
I think a set of function libraries is still important, for the same reason
that lots of REBOL's features are: people expect them. It gives them a
comfortable place to start, where they can look at things in a familiar
context. Dialects are where it's at, but people looking at REBOL for the
first time aren't going to see that. They need to be able to ease in slowly
in most cases.
There are also some basic pieces still missing (rounding numbers, justifying
strings, etc.) that could stand to have a good reference version in a
visible place. Other libraries would be higher level, more specialized (like
Ladislav's %high-fun.r), or OS specific. Dialects should probably have their
own area.
I think the current library suffers from lack of visibility and a bit of
stagnation, but it's still very valuable! For people just checking things
out on the web, they can't see how cool it is from the View Desktop, which
is a shame. One of the greatest things is how "alive" things can be. I.e.
you can click on a script and see it run. I think an EasyVID style front end
on a library could be *very* cool. Just click samples to run them, maybe
have a place to enter test code and data, all the docs are there as well.
Now where is all that extra time I thought I had... :)
--Gregg
[3/4] from: ingo::2b1::de at: 28-Sep-2002 8:53
Hi Gregg, and all,
Am Fre, 2002-09-27 um 23.41 schrieb Gregg Irwin:
> Hi Tim, et al
> <<
<<quoted lines omitted: 4>>
> I was thinking about this because some time ago I put out a feeler message
> to see if people were interested in organizing a set of function libraries.
I think what hinders the conception of a rebol standard library the
most, is the missing of a module system. As long as Rebol doesn't
include a module system, it will be hard to handle library.
Or, we should decide to use one of the many module handlers floating
around (or create one), and decide that this one will have to be used by
all library scripts.
Kind regards,
Ingo
[4/4] from: greggirwin::mindspring::com at: 28-Sep-2002 13:49
Re: Dialects (any plan is better than no plan) - Baneor Blessing?
Hi Ingo,
<< I think what hinders the conception of a rebol standard library the
most, is the missing of a module system. As long as Rebol doesn't
include a module system, it will be hard to handle library. >>
I agree completely. I thought RT was working on a module system for Core
3.0, so I kind of hesitate to start building something like a standard set
of libraries that would need to be redesigned when that happens.
<< Or, we should decide to use one of the many module handlers floating
around (or create one), and decide that this one will have to be used by
all library scripts. >>
Mabye the libmaster, when they are named, could talk with Carl about it and
get his recommendation, since he's "in the know" about what might be coming
down the pike. Then we might at least not be too far off track.
--Gregg
Notes
- Quoted lines have been omitted from some messages.
View the message alone to see the lines that have been omitted