Comments on: Host Kit for Linux released - More to follow

The R3 Host Kit allows developers to compile and link an R3 executable.

Yesterday the first Host Kit for Linux was released. Next, we'll work on OS X, but it doesn't use the .so shared lib model, so it may take more time to get it released.

Here's a summary of Host Kit releases:

59 Comments

Comments:

Brian Hawley 12-Dec-2009 15:00:09	An interpreter working on a non-jailbroken iPhone will be rejected by the App Store, so you won't be able to get it on the phone. There is no provision for user-installed applications - you have to go through the App Store, even for apps you write yourself. We would be better served by writing development tools in REBOL that generate source code for some other iPhone development platform, or to write an LLVM front end for a compilable REBOL dialect.
-pekr- 12-Dec-2009 15:17:09	Screw iPhone then. This is not OS, if it does not let you use whatever development system you want. Where's Windows mobile on the list? :-)
Brian Hawley 12-Dec-2009 15:57:05	Windows Mobile is definitely on my list of platforms that I need R3 support for. I'm already looking into porting the host code, time permitting. A build of r3-lib.dll for Windows Mobile ARM would be greatly appreciated - and probably x86 as well for testing in emulators, if x86 emulators are still used with recent releases of the OS and dev platform. I have a WM5 (for now) smartphone to test on, and can use the emulators of course.
-pekr- 13-Dec-2009 1:36:04	Just a note - Visual Studio 2005, 2008 support development environment for Windows Mobile applications. It uses emulators,IIRC. But - you have to have full VS, not just Express edition. More here: http://msdn.microsoft.com/cs-cz/windowsmobile/default%28en-us%29.aspx
Kaj 13-Dec-2009 10:39:09	Got it running last night on Syllable Desktop (our non-Linux system). :-) And on Syllable Server, of course, but that's a given, being Linux. Will report on R3 chat.
Giuseppe Chillemi 13-Dec-2009 17:21:15	My vote goes to Windows Mobile and Android. Too much useful to have REBOL there. It will be the base for a lot of developments and additions.
Alan Macleod 14-Dec-2009 8:22:28	And Palm WebOS?
Maarten 14-Dec-2009 13:47:19	The masterplan kicks in... I'd vote Android and 64 bit Linux...
-pekr- 14-Dec-2009 14:12:32	I vote for mobile devices ASAP - big market ...
xRatio 14-Dec-2009 15:12:49	All fiddling in the skies... What about a useable rebol 3 shell? What about local statics for example?? Almost impossible to write extremely often needed recursive functions without local statics. But not provided by REBOL. And -please- forget this -sorry- idiotic OOP. xRatio (same as Ratio)
-pekr- 14-Dec-2009 15:45:55	xRatio - you are off-topic as much as you can be here. This thread is devoted to R3 porting effort and clearly follows development plan. If you would care to properly follow R3 development, you would not call it "all fiddling in the skies". If you have any ideas of how to make things better, you can put your wishes in CureCode system, or discuss them with developers ...
xRatio 14-Dec-2009 16:25:18	-pekr-, I did not mean you personally. But see above, my deer. :-) As experienced developer I'm missing LOTS of BASIC features and functions - in R2 and R3 The scope of variables (by default private!! We treat privates as ERROR! ) is a mess. Nobody did and nobody CAN write real, complex applications with such a system. Yes, I like REBOL. But there must be done much more at the fundamental level. Local statics for example. You certainly know what that is? A local static? That's what I mean. Could write a book about all these issues. Cheers xRatio (aka Ratio)
Carl Sassenrath 14-Dec-2009 18:03:32	xRatio: R3 Chat is a good place to talk about any issues like that. There are many experts there. BTW, there are already many "complex" applications written in REBOL. Complex as in "powerful", not as in "bloated". So, the scope of variables works quite well once you understand that they're not lexically nor dynamically scoped, they are "definitionally scoped". This is a more powerful model than lexical scoping and less problematic than dynamic scoping (which isn't used in many languages any more.) Also, regarding local statics, that's true... so get on Chat and make suggestions. Note that literal series within functions are by default local static, and we also have Closures in R3. There was a suggestion earlier to make a small mod to closures to provide persistent local statics, but wasn't much feedback on it as I recall... so it was dropped.
Steeve 14-Dec-2009 21:56:44	Woooh ... Ratio, I know 2 or 3 ways to have static locals in a rebol function. It's strange because it seems that your knowledge about Rebol are weak, but you behave as if you had already done the trick.
Brian Hawley 15-Dec-2009 0:58:30	xRatio: Try funct/with for local static variables. But watch out: In any language with static local function variables, the use of those variables can create problems with recursion or multitasking. Useful, but only with care.
Jerry Tsai 15-Dec-2009 2:03:59	Just so you know. I am working for a big company. we deal with money like 50 billions USD everyday. I am developing a flow engine using R3 for my company. It's in the testing stage and will be online next week. Thanks to REBOL. It's REBOL that makes this complex system relatively easy to develop.
Jerry Tsai 15-Dec-2009 5:29:13	Correction! It should be 50 billions USD every YEAR
Kaj 15-Dec-2009 10:10:45	Same difference. ;-)
Maxim Olivier-Adlhoch 15-Dec-2009 22:16:34	jerry, what do you mean by a flow system?
Jerry Tsai 16-Dec-2009 2:17:24	Maxim, it's kind of like jBPM. What I am trying to say is that R3 is stable enough and is very helpful for me to develop a complex system.
xRatio 16-Dec-2009 16:08:34	(at) Steeve 14-Dec-2009 21:56:44 Please give me just ONE example. Just ONE of for your "2 or 3 ways" is enough. :-) xRatio
xRatio 16-Dec-2009 16:16:56	(at) Brian Hawley 15-Dec-2009 0:58:30 Hi Brian, for my people I completely rewrote large parts of the official REBOL docu (Rebol-core, word-defs etc), used throughout well defined terms and definitions, inserted useable, immediatly working examples etc. REBOL docu is in many respects just a mess. Saw nowhere things linke "funct/with". Could you please give me a simple working example how to use it? Thank you, xRatio btw: Garbage collect (GC, recycle) is not working at all. Hope this is known to the expert Rebolers. xRatio
xRatio 16-Dec-2009 16:20:33	(at) Carl Sassenrath 14-Dec-2009 18:03:32 Hi Carl, please give me a link. As mentioned earlier I cannot write in Altme Bug of Altme, Windoze, browser?! Do'nt really know. Cheers, xRatio
-pekr- 16-Dec-2009 17:06:04	Ratio - link to what? R3 Chat is integral part of R3 Console. Have you ever used R3? Because if you would do so, you would not ask where to find R3 Chat :-)
xRatio 16-Dec-2009 17:44:01	(at) -pekr- 16-Dec-2009 17:06:04 No. And you know and understand, why. ;-) When I see, that there is an improved function in R3 I import it to R2 ;-) Cheers, xRatio
xRatio 16-Dec-2009 17:47:46	Emails We tried to migrate with our Email program to REBOL. SET-MODE is normally ok. But used to set filedates of emails according to email dates does not work reliably. Sometimes ok, very often esp. with large or many emails not. We get hundrets of megabytes of emails. REBOL cannot handle SET-MODES correctly in these cases. A consequence of the not working GC ?! -------------------- PARSE-HEADER-DATE (used by REBOLs import-email) is not flexible enough since email dates are often not written as expected by PARSE-HEADER-DATE. Insofar we could write --after many hours of analyzing the problem-- a workaround. SET-MODES [modification-date] remains as a severe problem. xRatio
xRatio 16-Dec-2009 19:28:12	(at) Carl Sassenrath 14-Dec-2009 18:03:32 Hi Carl, please forget beginners in programming. They have no overview, no experience, always are chatting overblown nonsens. LOGIC is for those modern newbies a curious, foreign word. In short: They do not understand very much. There was a suggestion earlier to make a small mod to closures to provide persistent local statics, but wasn't much feedback on it as I recall... so it was dropped. Carl, professional developers urgently need local statics -- to store datas, in recursive functions etc. Simple example: You wrote an excellent(!) recursive function (Delete-dir). The question is: how to insert easily a simple over all counter into it? Except using privates or even global vars? Yes, we indeed have to set a recursive flag (option, "refinement"). This is difficult enough and works only from inside your function. But what about INDIRECT recursions???!! --- Impossible! I'm sure, you understand what I mean. --------------------- Carl, you are on the right way. Please do not make any wrong concessions!! Making things easy is one of the most difficult things at all. It's real science. REBOL's activation stack should be available for users. The error handling must be improved basically: First REBOL always should call a user defineable error function. "Blocks" should ALWAYS be evaluated (just use a sort of print for that, print works perfectly!) Last, NOT AT ALL LEAST!! Forget OOP and datatype CHAR !! Cheers, xRatio
Steeve 16-Dec-2009 21:59:58	xRatio, here you got, have fun... 1/ fun: context [ counter: 0 does [counter: couter + 1] ] 2/ fun: has [static][ static: [counter 0] static/counter: static/counter + 1 ] 3/ fun: funct/with [][counter: counter + 1] [counter: 0]
Steeve 16-Dec-2009 22:06:47	Sorry, i'm wrong for the first one, it should be 1/ context [counter: 0 set 'fun does [counter: counter + 1]] For the same price, i add a fourth form. 4/ fun: use [counter][counter: 0 does [counter: counter + 1]]
Brian Hawley 17-Dec-2009 2:06:29	xRatio, the docs need some work, mostly due to a shortage of people who are willing to work on them. Does this mean you are volunteering to write docs? As for porting functions from R3 to R2, this has mostly been done already in the R2/Forward project (though it could use an update). Look in R3 chat for details - that is where the source is. As for statics, though Carl may not remember the result of that discussion about closures, I do. That is why we have funct/with. Stop trying to sell us on the need to support something we already support - it's tiresome.
Brian Hawley 17-Dec-2009 2:23:58	As for your other requests: REBOL's activation stack should be available for users. - Security hole. The stack is available to those with the security permissions. The error handling must be improved basically: First REBOL always should call a user defineable error function. - Unnecessary, and almost always a bad idea. Most errors need to be handled locally. A global error handling function is only good for showing error messages when all appropriate error handling has failed. And that is a UI problem, easily handled in the UI code - perhaps with a hook to a user-provided error handler, if appropriate. "Blocks" should ALWAYS be evaluated - How presumptuous, to think that the interpreter would be able to know which dialect a block of code is supposed to be evaluated with, or for that matter that the block isn't just data. That is for the programmer to decide. Forget OOP and datatype CHAR !! - This argument is over: The benefits of both the char! type and occasional OOP have been demonstrated thoroughly, and the code is already written to support both. The rest of the stuff you are complaining about is even more off-topic than the rest, since it applies to R2, while this blog is about R3. Please ask somewhere else.
Henrik 17-Dec-2009 3:45:53	"Blocks" should ALWAYS be evaluated This is a good one! That basically destroys the entire foundation for REBOL in ways, I don't know how to count: Dialects, security, speed, memory, contexts, evaluation, anything. Would it even be possible to write and execute REBOL code? You really hit one of the most central design pieces of REBOL. I have no idea where you get this stuff from, but it can't be because you have an understanding for the language.
Luca 17-Dec-2009 6:16:04	Do not feed the troll. :-)
Janko M. 17-Dec-2009 7:55:57	I am almost sure that if you wrap iPhone REBOL app into an executable you can go with it to the appstore. For example www.stonetrip.com (is 3d engine where you code in lua and has numerous games for iPhone). Unity3d games (also for iphone) are coded in javascript on top of Mono. PhoneGap (html/js/css for front ; python and similar for the back) for making applications. etc..
Brian Hawley 17-Dec-2009 15:19:46	Janko, all of those platforms are compiled to native or LLVM code ahead of time (even the Javascript on Mono), or run in the browser. It is interpreters that are prohibited, other than the Javascript one in Safari. REBOL packaged into an app would not be allowed because the REBOL interpreter would have to be included. If an interpreter is included then the user will be able to run arbitrary code that hasn't been reviewed by the App Store gatekeepers, and they don't want that, hence the ban. The core semantics of REBOL would have to change to make it compilable. However, REBOL is an excellent language to write compilers in and it is quite possible to write compilable dialects in REBOL, even ones that resemble REBOL on a surface level well enough to run most REBOL code. Only hand-optimized or otherwise advanced REBOL code would have a problem; procedural-style code would be fine, for the most part. The resulting code that runs on the iPhone itself would not be REBOL code, though it could support the REBOL data model and thus interoperate with REBOL code running on other machines.
-pekr- 17-Dec-2009 16:31:27	Brian - I can't see why encapped app, even if it includes interpreter inside, should be a problem, if it does not allow to run any other app, than the one packed into exe together with the interpreter itself? I can understand Apple prohibits usage of REBOL "as-is", where you can "do" other scripts ...
Brian Hawley 18-Dec-2009 11:39:06	I didn't say the rules make sense, or are in any way a good idea, even for Apple, but that is what we're stuck with on that platform. Apple is being stupid.
xRatio 18-Dec-2009 23:28:47	(at) steeve, no answer to my questions? Your comments provide no help for me. Please try it with useable examples if you like and can. (at) Brian Hawley Rewrote just for internal usage large parts of REBOL docus while reading them, also to better understand them myself ;-). Old issue: Always better a third party writes the docus. The ACTIVATION STACK is needed in professional programs for debug and error handling. Regarding BLOCKS it seems you misundertand my question. We certainly want the values, not the variable names. Sometimes we get the values sometimes not. Cannot see any systematic here. So in the moment we evaluate ("reduce") all arrays manually. If the value is a function, this is a special REBOL issue mixing/identifying functions with variables. Should be resolvable. CHARs are the smallest elements of strings, so each char is a string. Also mentioned earlier: We cannot change to R3, so long there is no useable console. Did not get any answer to this crucial issue. (at) Henrik, earlier in extenso explained that variables and functions are totally different things. To mix or identify them (also done in OOP) is a basic logical error with desastrous consequences. ----------- Anyway. Thanks for your work on REBOL. I never saw a more efficient, but more difficult to use IT language. Even Assembly and C were and are for me much easier to understand and to use. Would it be a good idea to easily use C from inside a REBOL script? Is it possible? Is there a C interface? Did not see it so far. xRatio
Henrik 19-Dec-2009 7:12:53	Regarding BLOCKS it seems you misunderstand my question. We certainly want the values, not the variable names. If you evaluate a block unconditionally, you lose 75% of what REBOL is about. Not evaluating blocks like that is the whole point of bindings, contexts and dialects!, and that's why we have functions like REDUCE, COMPOSE, PARSE, GET, etc. Words are also not always variables. This is because words may not always evaluate to a specific value. The important thing about words is their meaning, which you determine through binding to values in a specific context. Equally important is that words may have no binding, thus no meaning. Depending on their binding, if any, when evaluated, you get a specific value back. This forms the absolute base of REBOL, the equivalent of bricks in a house. Sometimes we get the values sometimes not. Cannot see any systematic here. This is because you are completely ignoring how bindings and words work, an important principle that any good REBOLer must know: >> b: 5 == 5 >> a: [b] == [b] >> a/1 == b ; word! >> get a/1 == 5 ; value >> c: context [b: 7] == make object! [b: 7] >> bind a c ; changing the context of 'a >> a/1 == b ; word! again >> get a/1 == 7 ; new value because of different binding Also know how to build and reduce blocks using lit-words: >> [b c 7] == [b c 7] >> ['b c 7] == ['b c 7] >> b: 3 == 3 >> c: 15 == 15 >> reduce ['b c 7] == [b 15 7] >> reduce reduce ['b c 7] == [3 15 7] Also mentioned earlier: We cannot change to R3, so long there is no useable console. Did not get any answer to this crucial issue. The host kit that you call "fiddling in the sky" forms the basis for building a proper console for each OS platform using open source, rather than the current method in R2, where only the Windows version has a proper console. CHARs are the smallest elements of strings, so each char is a string. Chars are not series, thus not strings. They eat less memory and make REBOL go faster. Henrik, earlier in extenso explained that variables and functions are totally different things. To mix or identify them (also done in OOP) is a basic logical error with desastrous consequences. No, functions are special contexts with input parameters that undergo the same conditions for binding as mentioned above. I'll tell you a little secret: Objects in REBOL are contexts. In R3, this fact is even more important, due to added security, closures, modules and tasks. I never saw a more efficient, but more difficult to use IT language. I can understand this if you come from C or assembly, but please try to learn that REBOL is very different from C or assembly with very good reasons.
batman 20-Dec-2009 9:15:04	xratio: would you be able to share your documentations with the rebol community? Henrik: would you be able to provide a link to the best document covering binding and context ? thanks in advance to both of you . cheers
Henrik 20-Dec-2009 10:11:22	Batman, Bindology is a good read, if a little technical: http://www.rebol.net/wiki/Bindology
Brian Hawley 20-Dec-2009 16:50:18	"The ACTIVATION STACK is needed in professional programs for debug and error handling." The security permissions that I mentioned are required are the debug permissions. Error handling works just fine even without these permissions, and includes a function call trace. No, neither of these can be ported to R2. We already have this feature. Stop requesting features that we already have once we've already said so - it gets tiresome. You aren't going to be able to understand what capabilities R3 offers already if you continue to refuse to use it. Not everything can be backported.
xRatio 20-Dec-2009 19:10:38	(at) Henrik 19-Dec-2009 7:12:53 Please don't tell us the same -sorry- nonsense again and again. Do you remember and still know that you even are not able to understand basic functions like SUBSTR ? Just an example. xRatio
xRatio 20-Dec-2009 19:14:16	(at) batman 20-Dec-2009 9:15:04 xratio: would you be able to share your documentations with the rebol community? In the moment my docu is for internal use, concentrated on most vital things including VID. The very best publicly available external REBOL docus are created by Nick Antonaccio. - In my opinion, after seeing all others, the best. Cheers, xRatio
-pekr- 21-Dec-2009 0:51:50	(at)xRation - the only -sorry- nonsese in this thread comes from you. You just managed to totally ruin the thread. I can tell you, that if it would be a forum, you would be banned from the thread for being off topic, for your rants and attitude. You claim yourself being a professional developer, but I can tell you, from what you write - even me, as a very average REBOL coder - can recognise, you really lack deep REBOL knowledge. I think that guys here try to be really helpfull, so you should be a bit more respectfull to them. Also - start to use R3, and stop your excuses for not being able to do so. As for SUBSTR - what's there to not understand it? There were several discussions about if we should add such functions as left, right, substr, etc. into REBOL. The person properly udnerstanding how REBOL series work, either does not need them, or writes them as one-liner functions and adds them into their library, and that's it ... Once again - this thread is about R3 Host kit - stop missusing it with nonsense. If you want to chat, just manage to get yourself to AltME, R3 Chat, or mailing list, easy as that. Hundreds of ppl are able to understand it, you should follow them :-) ...
Henrik 21-Dec-2009 8:33:47	Please don't tell us the same -sorry- nonsense again and again. "Us"? I'm only telling you, because you are the first I've seen to persistently complain about this in the 12 years that REBOL has existed, by persistently denying to understand the basics of REBOL, and when discarding 30 years of thinking about programming language design as "nonsense", what's there to do? So you don't recognize that REBOL has concepts like contexts, bindings and dialects? Why is noone else jumping in and correcting me, when I spend time writing examples for you? I happen to do that, because I enjoy teaching REBOL. But, it's good that you show your stance in public, because based on your status as a "professional developer", you are not likely to be hired for professional REBOL work, until you start over and learn REBOL from scratch. If someone has, well... good luck. :-)
Nick 21-Dec-2009 8:55:07	Batman, http://blog.revolucent.net/2009/07/deep-rebol-bindology.html has some good info about contexts, which should make the whole bind topic a little easier to get.
Nick 21-Dec-2009 9:49	xRatio, thank you for the compliment, but my take on this conversation is in line with others who've responded here. There are some fundamental concepts at the heart of REBOL that make it so simple, powerful, and productive. Your point about substrings is a great example. REBOL can't be used effectively without understanding series, so for REBOLers there's simply no need to implement the substring function you've requested. If you're a whiz with series, you'll think about it in those terms. What others are suggesting is that you embrace the REBOL approach - in the case of substrings, use series functions - because that approach will apply to solving many other problems, simply and consistently, with REBOL. The same is true for other issues that make REBOL unique. REBOL does require relearning some familiar mainstream coding techniques, but REBOLers prefer the REBOL approach as a more productive solution, so you'll tend to find that others don't think it should be changed. The point is not to argue with you but to help you understand some more of the practical benefits of REBOL's design :)
xRatio 22-Dec-2009 18:48:58	(at) -pekr- 21-Dec-2009 0:51:50 Please THINK before you write personally aggressive nonsens. If you have a solution for SUBSTR - TELL US. ;-) (at)Nick 21-Dec-2009 9:49 Agree, but we never should forget the rules of logic!!! (at)Henrik, you write just nonsens. Sorry. xRatio
xRatio 22-Dec-2009 18:50:51	Yes, I'm one of the rare professionals, successfully working with CPUs since decades. Almost immediately seeing what's going wrong (also in REBOL). Bloody beginners should never be admitted to the roots. xRatio
Ashley 22-Dec-2009 18:54:24	I'd go further and say it's a different way of thinking. Some people get it, some people don't. Recently, after showing someone a snippet of REBOL code that did in 3 lines what they took almost 20 to do in VB, they exclaimed, "but HOW can you think like that", to which I could only reply, "but how can you NOT think like that". It's very natural for me [now] to express in a line of readable REBOL code what other languages would force you to do in multiple steps/lines ... and then comment for good measure.
Ashley 22-Dec-2009 19:05:21	"If you have a solution for SUBSTR" substr: func [string [string!] pos [integer!] length [integer!]] [ copy/part at string pos length ] Simple.
Steeve 22-Dec-2009 23:09:48	Ashley, It can't be so simple. You have cheating or something. Because I know well an expert, successfully working with CPUs since decades immediatly seeing what's going wrong (also in REBOL), who was unable to find it. So, it must be a complicated solution or nothing.
batman 22-Dec-2009 23:50:49	Henrik: & Nick: thanks for the link because of my beginner level, I must admit the http://blog.revolucent.net/2009/07/deep-rebol-bindology.html link was easier to digest than the http://www.rebol.net/wiki/Bindology I am just wondering since this is a very important concept, why don't we link them directly to the rebol documentation web page. cheers batman
xRatio 23-Dec-2009 22:18:20	(at) Ashley, of course, all is "simple" if you have a solution. Do you imagine how long new users must search for it??? SUBSTR is a basic function, must be available always. It's just an example..... xRatio
Henrik 24-Dec-2009 4:21:52	Do you imagine how long new users must search for it??? I know, learning (and googling) is a bitch. Nick, great link!
amused 24-Dec-2009 11:55:49	this is the first time i visit this site, when i was following a link from a blog. at xratio (at) Ashley, of course, all is \"simple\" if you have a solution. please forget beginners in programming. They have no overview, no experience, always are chatting overblown nonsens. Yes, I\'m one of the rare professionals, successfully working with CPUs since decades. Almost immediately seeing what\'s going wrong (also in REBOL). Bloody beginners should never be admitted to the roots. Isn\'t that an irony? How can you one of the \"rare\" \"professionals\" with decades of successful experience and \"almost immediately seeing what\'s going wrong\" not be able to come up with that simple solution ? or at least do a simple search through google. Professional? are you serious ? or just a programmer? do you know that true professionals abide by a code of conduct ? For one, to not use foul language like \"bloody beginners\" when one is participating in any civilized discussion For two, not to look down at and be condescending towards beginners . True Professionals always try to coach/teach beginners. did you forget that at one point you were a beginner ? anyway please note I won\'t be coming back to this site, I came here only because I was reading a blog at www.batman12345.wordpress.com and there was a link to here.
Hostile Fork 26-Dec-2009 13:30:44	Instead of people debating on Carl's blog (when the topic is clearly the Host Kit) could everyone please migrate alternative discussions and proposals to The Wiki? Rather than respond here I've started a new page to discuss Ratio's concern about substring. Here's the talk page for Substring In Rebol.
AlexC 9-Jan-2010 3:35:05	I'm hoping the pending Amiga version will come out as well. While a .library would be the most efficient, since AmigaOS4.1 supports shared objects too maybe it will ease supporting that platform.

Post a Comment:

You can post a comment here. Keep it on-topic.