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

ANN: Rugby 43 experimental

 [1/7] from: koopmans:itr:ing:nl at: 31-Oct-2001 10:37


Hi, I am pleased to announce the release of Rugby 4.3 experimental AFAIK it is stable, but it will have status experimental until the docs are finished. You can download it at: http://www.rebolforces.com/~erebol/rugby43e.tar.gz What's new: 1) Transparent error propagation. Errors are propagated back as error! instead of object! Thanks to Ladislav and Cyphre 2) Support for function refinements. YES! You can use refinements on functions (securely). Thanks to joel neely for the use [ myself ] [ myself: func ...] trick 3) Improved: stub generation. Needed for the function refinements. It is no longer necessary to put things in a context when using multiple rugby services. Unless they define the same functions, of course. Rugby now has one of the hardest parts of meta code I have ever seen, in server.r (in the modules directory). Look at the build-proxy func and try to figure it out all yee Rebol wizards ;-) If you find bugs, please report them. Enjoy and thanks, Maarten Koopmans

 [2/7] from: koopmans:itr:ing:nl at: 31-Oct-2001 13:15


And added the first minor bugfix! --Maarten On Wednesday 31 October 2001 10:37, you wrote:

 [3/7] from: petr:krenzelok:trz:cz at: 31-Oct-2001 14:50


Hi Maarten, so I looked at Rugby for few mins and - - I was able to set it up (well, there is no setup at all :-) and get it running in 1 minute ... now few questions though: - I was interested in ability to get some kind of help or examination, upon what server provides me. Is there any such functionality available? (e.g. simply returning env serve block? - so I imported functions from server, to have them "locally" (by using get-rugby-service), and was surprised by "bloat" I got, by probinx resulting object :-) I thought that I will get real source code binded locally? But maybe then it would not be secure at all, to know what is server doing on the machine it is running on :-) Well, then once again - will there be any possibility to get built-in help on remote functions? - it seems very slow here. I was surpised by its slowness. Am I doing anything wrong?: server-side .... ->> test1: does ["nothing"] ->> server [test1] client-side .... ->> ble: context get-rugby-service tcp://localhost:8001 ->> start: now/time for i 1 100 1 [ble/test1] print now/time - start 0:00:44 44 sec on my P300, locally, to get simple echo, 100 times? I would rather expect 44 request per sec :-) So? Thanks, -pekr- Maarten Koopmans wrote:

 [4/7] from: koopmans:itr:ing:nl at: 31-Oct-2001 15:45


Hi, On my celeron 500 laptop: start: now/time loop 100 [ echo "la"] print (now/time - start) ==10 secs. Note that the client stub code (which is what you are testing is extremely dynamic as of this release). But as the load is on the server.... you should test it in a multiple machine setup if you want to do performance tests. The stub code you get back does automatic rexec and sexec on the right machines with refinement passing etc. Hence it is very dynamic code, and it may look bloated. Especially the refinement transferral from the stub to the other side was tricky. If you put the result of get-rugby-service in a context o, all values except self of first o will be the ones that you can get from this service. I'll put that in the docs. Try help on any function you have imported (although you'd have to do the import in the global context because of help), it will work. HTH, Maarten

 [5/7] from: petr:krenzelok:trz:cz at: 31-Oct-2001 16:43


Maarten Koopmans wrote:
> Hi, > On my celeron 500 laptop:
<<quoted lines omitted: 3>>
> dynamic as of this release). But as the load is on the server.... you should > test it in a multiple machine setup if you want to do performance tests.
Why? Does it matter if load is on the server? I mean - how does it help? I think that local machine test is the fastest communication possible? Imagine chat program, e.g. goRim, where e.g. 10 clients are connected. I type really fast or so my colleagues tell me :-) The mechanism doesn't seem to be fast enough even to give me real-time feeling (tested just simple single echo in console). Or I just missunderstand something important here ... -pekr-

 [6/7] from: koopmans:itr:ing:nl at: 31-Oct-2001 17:09


On one machine you are switching between processes which on a reasonble loaded machine may be slower. On the other hand... I am not seeing the same thing here.... so.... And if you have played with gorim you know that there really is no problem (yet ;-) Graham, I am curious what an upgrade to 43e would do to gorim. I guess (or hope) nothing. --Maarten On Wednesday 31 October 2001 16:43, you wrote:

 [7/7] from: gchiu:compkarori at: 1-Nov-2001 12:20


> Graham, I am curious what an upgrade to 43e would do to > gorim. I guess (or > hope) nothing. >
Me too :) I'm in the process of shifting the server to xrebol.net and might try it this weekend. -- Graham Chiu

Notes
  • Quoted lines have been omitted from some messages.
    View the message alone to see the lines that have been omitted