Mailing List Archive: 49091 messages
  • Home
  • Script library
  • AltME Archive
  • Mailing list
  • Articles Index
  • Site search
 

[REBOL] Re: View for CE or Zaurus revisited.

From: petr::krenzelok::trz::cz at: 13-Feb-2003 22:53

Brian Hawley wrote:
> At 02:08 PM 2/12/03 -0800, TBrownell wrote: > >> Why oh why is there no Windows CE/view product? > > Memory and speed requirements. View typically uses at least > 4 to 5 mb of memory for every instance, and that's before > you take any data or scripts into account. Also, /Core on > CE is dog slow (at least on ARM, I haven't tested others). > Both of these problems together make REBOL impractical for > any embedded systems platform, even a generous one like CE.
Yes, very strange thing. Carl designed AmigaOS - modular concept, fast enough even on 7MHz machine, yet View is sometimes laggy even on pretty decent machine. Rebol is very good concept, but inability to compile (or overall lack of free time of RT to add domain specific dialect virtual machines) does not provide us with general enough language - we are magnitude slower than C or even Java on general math and GUI stuff (general primitives, bitmap handling). Maybe we should wait for next gen devices? The question is - is e.g. iPAQ (or other) PDA, running XScale at 400 MHz, 64 MB fast enough? If so, maybe RT could think about their "next thing" as mobile devices support - Symbian + PocketPC + Linux? I think that Ed said it very well! On nearly month basis we see various products popping up here or there, which could be done using rebol ... nearly done. We should stop playing the lowest common denominator game, that is plain silly. In the long term, we are killing rebol oportunities in certain areas. We need more open aproach and no that pseudo efficiency game - not adding usefull features just because they would add to executable size - the rest of world does not care - the rest of world will not judge rebol upon its executable size, but what can be done using rebol - and that is the only thing we should keep in mind (and that does not mean we should forgot efficiency) ...
> Fixing the memory problem would require reorganizing REBOL > to share more resources between instances. The speed issue > would require a lot of profiling and hand optimization. > Both together might require almost a separate code base,
maybe few things could be done even today ... during czech rebol workshop Ladislav told us few hints that could possibly help (GC and memory allocation related), but only Carl could answer those ones ...
> I can dream, right?
Who knows? The last dev. update is dated November, we have yet to see, what RT plans for 2K3. So far we know that RT is working on OsX ports, SDK is still beta ... I just hope RT will not be silent half a year ... So, in the meantime, we can forget multimedia and other cute things Rebol could shine in, and code few boring enterprise apps, can't we? :-) There is always something usefull to do with Rebol ... -pekr-