Parallel Ant (was RE: The Complete Server Platform?)
From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED]] Cannot ant (like normal decent pmake/bsdmake) figure out from the dependencies what can be done in parallel. I am not asking for the awsomeness of 'make -j 8 world' of *BSD - butsomething close should be possible I take it - could be a nice graduade student project :-) It's probably feasible, but hard, and arguably the wrong place to do it. I assume that the time consuming part is the javac. If there's another task that's taking time, then it might be worth looking at, but for javac... Ant gets most of its speed gains (over make) by passing all the java code to javac at once. If it were passing the files one at a time, then it would be relatively easy to run two processes at once. But doing that would slow you down in almost all cases (maybe not on the 8-way box, but in 99% of cases it would) So Ant would have to take the fileset you pass, and do dependency analysis on it to produce two (or more) disjoint sets of files to pass separately to javac. There's a fair amount of complexity there, and I'm not sure that Ant is the place to put it. Since the normal cases for JavaC is to pass in multiple files at once, I think javac should be able to parallelise itself. So, a better grad project, might be to hack Jikes to support parallel compiles. [ I think follow-ups should go to Ant-Dev ] -- NOTICE This e-mail and any attachments are confidential and may contain copyright material of Macquarie Bank or third parties. If you are not the intended recipient of this email you should not read, print, re-transmit, store or act in reliance on this e-mail or any attachments, and should destroy all copies of them. Macquarie Bank does not guarantee the integrity of any emails or any attached files. The views or opinions expressed are the author's own and may not reflect the views or opinions of Macquarie Bank. -- To unsubscribe, e-mail: mailto:[EMAIL PROTECTED] For additional commands, e-mail: mailto:[EMAIL PROTECTED]
Re: The Complete Server Platform?
On 3/22/02 2:59 AM, Stefan Bodewig [EMAIL PROTECTED] wrote: On 21 Mar 2002, Jason van Zyl [EMAIL PROTECTED] wrote: I assume that ant is not made to take advantage of a multi-processor box, Ant isn't doing too many things that could take advantage of multiple processors - it doesn't compile itself but uses your JDK's javac (you know that 8-) which won't take advantage of multiple processors for example. If there are things in your build process that can be done in parallel, you can use Ant's parallel task (Ant = 1.4) and run them in parallel. This should take advantage of multiple processors if your JVM uses native threads. If you put javac inside parallel, make sure you fork new processes though as Sun's javac code doesn't seem to be thread-safe. Woo hoo! This is exactly what I was going to do to ant - I have a dual proc mac, and wanted my builds to go even faster... -- Geir Magnusson Jr. [EMAIL PROTECTED] System and Software Consulting You're going to end up getting pissed at your software anyway, so you might as well not pay for it. Try Open Source. -- To unsubscribe, e-mail: mailto:[EMAIL PROTECTED] For additional commands, e-mail: mailto:[EMAIL PROTECTED]
Re: The Complete Server Platform?
Jason van Zyl wrote: I assume that ant is not made to take advantage of a multi-processor box, (I compiled some code on a quad processor machine and ant didn't really seem to move that much faster then on my laptop) but if I compile ant using gjc would it take advantage of a multi-processor machine? It would be no different. Sun's JRE and gcj can both take advantage of multiprocessor machines when threading is in use. If threading isn't in use you will only utilise one processor. -- Pete -- To unsubscribe, e-mail: mailto:[EMAIL PROTECTED] For additional commands, e-mail: mailto:[EMAIL PROTECTED]
Re: The Complete Server Platform?
On Fri, 2002-03-22 at 02:59, Stefan Bodewig wrote: On 21 Mar 2002, Jason van Zyl [EMAIL PROTECTED] wrote: I assume that ant is not made to take advantage of a multi-processor box, Ant isn't doing too many things that could take advantage of multiple processors - it doesn't compile itself but uses your JDK's javac (you know that 8-) which won't take advantage of multiple processors for example. If there are things in your build process that can be done in parallel, you can use Ant's parallel task (Ant = 1.4) and run them in parallel. This should take advantage of multiple processors if your JVM uses native threads. If you put javac inside parallel, make sure you fork new processes though as Sun's javac code doesn't seem to be thread-safe. Are there any solutions that would make a 4-way machine appear as a machine with a single processor so that single threaded code can take advantage of those extra processors transparently? I have a 4-way box here that is sitting idle here @ zenplex and I may have access to 2 8-way machines but if running ant isn't going to be able to take advantage of these machines (in some form, don't mind doing some work) then I might as well run builds on my build box at home. Stefan -- To unsubscribe, e-mail: mailto:[EMAIL PROTECTED] For additional commands, e-mail: mailto:[EMAIL PROTECTED] -- jvz. Jason van Zyl [EMAIL PROTECTED] http://tambora.zenplex.org -- To unsubscribe, e-mail: mailto:[EMAIL PROTECTED] For additional commands, e-mail: mailto:[EMAIL PROTECTED]
Re: The Complete Server Platform?
Jason van Zyl [EMAIL PROTECTED] writes: On Thu, 2002-03-21 at 15:48, Jon Scott Stevens wrote: on 3/21/02 8:41 AM, Jason van Zyl [EMAIL PROTECTED] wrote: I assume that ant is not made to take advantage of a multi-processor box, (I compiled some code on a quad processor machine and ant didn't really seem to move that much faster then on my laptop) but if I compile ant using gjc would it take advantage of a multi-processor machine? How would what is currently a shell script that sets some variables and executes a JVM be made to take advantage of a multi-processor box? I assumed Costin compiled the java goods. Just because you have multiple processors, it doesn't mean that the load for one single process is going to be distributed across them. It would take Ant to do all sorts of parallel threading to even get close to making that happen and since Ant is really a single threaded application, I doubt you will ever see a speed up (native code or JVM code). I was looking for some magic :-) make -j3 -- To unsubscribe, e-mail: mailto:[EMAIL PROTECTED] For additional commands, e-mail: mailto:[EMAIL PROTECTED]
Re: The Complete Server Platform?
On Fri, 22 Mar 2002, Geir Magnusson Jr. wrote: On 3/22/02 2:59 AM, Stefan Bodewig [EMAIL PROTECTED] wrote: On 21 Mar 2002, Jason van Zyl [EMAIL PROTECTED] wrote: I assume that ant is not made to take advantage of a multi-processor box, Ant isn't doing too many things that could take advantage of multiple processors - it doesn't compile itself but uses your JDK's javac (you know that 8-) which won't take advantage of multiple processors for example. If there are things in your build process that can be done in parallel, you can use Ant's parallel task (Ant = 1.4) and run them in parallel. This should take advantage of multiple processors if your JVM uses native threads. If you put javac inside parallel, make sure you fork new processes though as Sun's javac code doesn't seem to be thread-safe. Woo hoo! This is exactly what I was going to do to ant - I have a dual proc mac, and wanted my builds to go even faster... Cannot ant (like normal decent pmake/bsdmake) figure out from the dependencies what can be done in parallel. I am not asking for the awsomeness of 'make -j 8 world' of *BSD - butsomething close should be possible I take it - could be a nice graduade student project :-) DW -- To unsubscribe, e-mail: mailto:[EMAIL PROTECTED] For additional commands, e-mail: mailto:[EMAIL PROTECTED]
Re: The Complete Server Platform?
On Thu, 2002-03-21 at 15:48, Jon Scott Stevens wrote: on 3/21/02 8:41 AM, Jason van Zyl [EMAIL PROTECTED] wrote: I assume that ant is not made to take advantage of a multi-processor box, (I compiled some code on a quad processor machine and ant didn't really seem to move that much faster then on my laptop) but if I compile ant using gjc would it take advantage of a multi-processor machine? How would what is currently a shell script that sets some variables and executes a JVM be made to take advantage of a multi-processor box? I assumed Costin compiled the java goods. Just because you have multiple processors, it doesn't mean that the load for one single process is going to be distributed across them. It would take Ant to do all sorts of parallel threading to even get close to making that happen and since Ant is really a single threaded application, I doubt you will ever see a speed up (native code or JVM code). I was looking for some magic :-) -jon -- To unsubscribe, e-mail: mailto:[EMAIL PROTECTED] For additional commands, e-mail: mailto:[EMAIL PROTECTED] -- jvz. Jason van Zyl [EMAIL PROTECTED] http://tambora.zenplex.org -- To unsubscribe, e-mail: mailto:[EMAIL PROTECTED] For additional commands, e-mail: mailto:[EMAIL PROTECTED]
Re: The Complete Server Platform?
On 21 Mar 2002, Jason van Zyl wrote: ( the startup time is just amazing, you'll not realize you run 'ant' instead of 'ls' ) I assume that ant is not made to take advantage of a multi-processor box, (I compiled some code on a quad processor machine and ant didn't really seem to move that much faster then on my laptop) but if I compile ant using gjc would it take advantage of a multi-processor machine? No - ant doesn't have too much overhead itself, it's what it executes. Compiling ant to native reduced the startup time to almost zero - that's the time it takes to load the huge VM, hundreds of classes, JIT-compile them. About 3-4 seconds per invocation. All this goes away ( except the first time you run them ). Plus a lot of stuff that is not used or is used only once will remain on disk and not take up RAM ( thanks to the OS paging of executables ). If you use ant with javac - again you may be able to cut all the loading/jit-ing of javac - but I wasn't able to compile javac to native ( I haven't tried actually - I used jikes ). It's kind of similar with jikes vs. javac. Regarding multiprocessor - ant is singlethreaded, but if you fork the compilers it'll take advantage of the other processors. ( jikes or any native compiler will most likely be scheduled to a different process ) Costin -- To unsubscribe, e-mail: mailto:[EMAIL PROTECTED] For additional commands, e-mail: mailto:[EMAIL PROTECTED]
Re: The Complete Server Platform?
Daniel Rall wrote: Also, would you point me to a reference on how memory management is handled? It uses the Boehm collector. For full details see here: http://www.hpl.hp.com/personal/Hans_Boehm/gc/ It is a conservative collector that can also implement garbage collection for C and C++, though gcc doesn't yet take advantage of this. It's a fairly decent collector, I think, that implements most of the obvious optimisations such as generations, exploitation of the MMU, etc. Its conservative nature is both an upside and a downside. On the downside, a small amount of storage may not get reclaimed -- though unlike leaky C programs, this amount tends not to increase with time. On the upside, there is less overhead because the collector doesn't have to distinguish between pointers and other types. -- Pete -- To unsubscribe, e-mail: mailto:[EMAIL PROTECTED] For additional commands, e-mail: mailto:[EMAIL PROTECTED]
Re: The Complete Server Platform?
On Tue, 19 Mar 2002, Jon Scott Stevens wrote: One other question: Is there a valuable performance enhancement to compiling to native code with gcj? Right now - no, I couldn't notice any significant difference while running tomcat. It is as fast as IBM JIT ( and faster than hotspot ). However it start much faster, and get to the peak performance sooner ( since all optimizations are already done ). Another nice thing is that all the java code is actually compiled to .so - and fully interoperable with C++ ( for C you need to deal with the mangling ). That's very similar with C# and .net languages, and it's not bad at all. GCJ can be a very good answer to .NET, with a bit of work it may allow the same language independence ( GCC already does a lot in this direction ). Costin -- To unsubscribe, e-mail: mailto:[EMAIL PROTECTED] For additional commands, e-mail: mailto:[EMAIL PROTECTED]
Re: The Complete Server Platform?
Daniel Rall wrote: Does the bytecode interpreter [from gcj] handle class loading yet? Yes. You can invoke the bytecode interpreter directly with gij if you don't want to compile. Gij will handle Class.forName and friends correctly. If you compile to native code, the resulting executable will first of all search shared libraries for the class you ask for. If it doesn't find it, it will search for a bytecode file that it can interpret. In other words you can freely mix interpreted and compiled code. -- Pete -- To unsubscribe, e-mail: mailto:[EMAIL PROTECTED] For additional commands, e-mail: mailto:[EMAIL PROTECTED]
Re: The Complete Server Platform?
on 3/19/02 8:36 AM, Daniel Rall [EMAIL PROTECTED] wrote: Pete Chown [EMAIL PROTECTED] writes: Daniel Rall wrote: Does the bytecode interpreter [from gcj] handle class loading yet? Yes. You can invoke the bytecode interpreter directly with gij if you don't want to compile. Gij will handle Class.forName and friends correctly. If you compile to native code, the resulting executable will first of all search shared libraries for the class you ask for. If it doesn't find it, it will search for a bytecode file that it can interpret. In other words you can freely mix interpreted and compiled code. Thanks Pete, very informative. - Dan One other question: Is there a valuable performance enhancement to compiling to native code with gcj? -jon -- To unsubscribe, e-mail: mailto:[EMAIL PROTECTED] For additional commands, e-mail: mailto:[EMAIL PROTECTED]
Re: The Complete Server Platform?
Pete Chown [EMAIL PROTECTED] writes: Daniel Rall wrote: Does the bytecode interpreter [from gcj] handle class loading yet? Yes. You can invoke the bytecode interpreter directly with gij if you don't want to compile. Gij will handle Class.forName and friends correctly. If you compile to native code, the resulting executable will first of all search shared libraries for the class you ask for. If it doesn't find it, it will search for a bytecode file that it can interpret. In other words you can freely mix interpreted and compiled code. Thanks Pete, very informative. - Dan -- To unsubscribe, e-mail: mailto:[EMAIL PROTECTED] For additional commands, e-mail: mailto:[EMAIL PROTECTED]
Re: The Complete Server Platform?
On 19 Mar 2002, Pete Chown wrote: Daniel Rall wrote: Does the bytecode interpreter [from gcj] handle class loading yet? Yes. You can invoke the bytecode interpreter directly with gij if you don't want to compile. Gij will handle Class.forName and friends correctly. I actually tested GCJ ( in 'native mode' ) with ant, tomcat(3), crimson ( required by both ). It had some minor issues ( required few changes, inner classes are tricky even for jikes ), and few bugs in class loading - but I was able to get everything working. The good/bad news - it's was as fast as IBM JDK1.3 on linux ( which is the fastest - on my setup ). Beeing as fast as the fastest is usually good news, but I was hoping a bit more :-) ( the startup time is just amazing, you'll not realize you run 'ant' instead of 'ls' ) Costin -- To unsubscribe, e-mail: mailto:[EMAIL PROTECTED] For additional commands, e-mail: mailto:[EMAIL PROTECTED]
Re: The Complete Server Platform?
Pete Chown [EMAIL PROTECTED] writes: The other thing I would like to push is gcj. It doesn't seem to be very well known. For people who haven't come across it, it is part of gcc and it is an ahead-of-time compiler for Java. It also includes a bytecode interpreter so it can deal with dynamically generated code, and a free implementation of the Java class libraries. Does the bytecode interpreter handle class loading yet? -- To unsubscribe, e-mail: mailto:[EMAIL PROTECTED] For additional commands, e-mail: mailto:[EMAIL PROTECTED]
RE: The Complete Server Platform?
'kay. Summary: (everyone, please correct and add to?) --- SIMILAR EFFORTS --- Here's a list of similar efforts that I know of... (requirments: 1) some kind of application platform 2) 100% java 3) open source) JBoss - goal:provide open source J2EE server. offers: J2EE compliance, EJB, JMS, O/R-mapping, JNDI, JCA, JTA/JTS, JAAS, GUI Management (JMX), Servlet/JSP, Database, JDO license: LGPL url: http://www.jboss.org status: release notes: huge developer community Jahia - goal:provide portal solution built on J2EE components. offers: JBoss, Database, Servlet/JSP, custom portal server (dunno what it does as the installer fails on my win2000 box) license: Jahia License (commercial) url: http://products.xo3.com:8080/ status: (preliminary) release notes: Paul, seems like a shrink-wrap commercial solution to me??? Enhydra --- goal:provide open source web application server. offers: Servlets/JSP, Template Engine (XMLC), Sessions, Database connectivity, MVC Framework license: Enhydra Public License (Mozilla-style) url: http://www.enhydra.org status: release note:dying slowly it seems, as Lutris is pulling out EAS --- goal:provide BSD-style license J2EE server. offers: Servlets/JSP, EJB, Avalon license: BSD-Style url: http://eas.betaversion.org status: alpha note:been dead a while now as backing company went bankrupt. OPEN ENTERPRISE DISTRIBUTION goal:provide an open source alternative architecture to J2EE. offers: Database, Pooling, Logging, Management, etc. (to be defined) license: Apache-style url: none (will be sourceforge) status: conceptual Candidate Components For Inclusion -- Jakarta: Avalon Struts Turbine Velocity James Jetspeed Slide Tomcat Apache XML: Cocoon SOAP SourceForge: Enterprise Object Broker Simper (soon, anyway) HSQL Ozone JBoss Maverick Jetty Niggle OpenCMS Pi (...the list goes on and on...) Exolab: OpenEJB OpenJMS OpenORB Castor Tyrex (...) Clearly, there are loads. We need some criteria. Architecture One thing I think OEB needs is formalization of the patterns of choice in the form of core interfaces. The two projects I know that deal with that are Avalon and Turbine. Another thing it needs is a definition of how it glues. Options are JNDI, SOAP, JMX, Avalon Service framework, Turbine Services Framework, CORBA...and I'm sure I'm missing a few. Of those, of course _I_ am in favor of the setup Avalon uses, but hey ;) This e-mail is now officially more than long enough. cheers, - Leo -Oorspronkelijk bericht- Van: Andrus Adamchik [mailto:[EMAIL PROTECTED]] Verzonden: maandag 25 februari 2002 1:27 Aan: Jakarta General List Onderwerp: Re: The Complete Server Platform? Tim Hyde wrote: Andrus, I'm 100% behind the idea of the complete platform, but I'm worried that your proposal talks about 'Web Applications'. I believe that what's needed is an alternative to the very idea that J2EE (or even J2SE) is *the* definitive collection of java libraries, and that the project should offering a number of sensible alternatives for use in any architecture. We still will depend on certain commercial JVM's (Sun or IBM), right? Database access, Logging, and Development Process are three things that you've specifically mentioned that aren't particularly Web or Server oriented. Web Applications, or Server Applications, are more part of today's 'fashion' than inherent categories of how you make a computing solution, and we can expect things to move on during the lifetime of Java. Well, we can hope, anyway. :-) You are right. I was mentioning Web applications just cause I wanted to limit initial scope to something sane. And I guess because I am myself is a better expert in this area then in any other. This would've helped to concentrate on a certain solution-based approach from the beginning. But I agree we can widen the scope as long as we can outline the problems being solved. So, if possible, why not talk about a 'development and deployment platform for Java applications' - and then start off by assembling both the underlying 'component' toolsets and a number of combination-examples, such as the jGuru one Ted mentioned, and whatever else might emerge during the project as perhaps 'miniature live examples'. +1, like I said above, I am for it if we define use cases we are going after. Naturally, server applications are the primary interest point initially, but it would be nice to think that the collection of tools being provided for distribution would be offered
RE: The Complete Server Platform?
One small extra: if a RedHat style toolkit distribution were available, the number of independent consultants who could offer their support services would exceed the number available to BEA, for example, eliminating that argument that 'I buy where I can depend on getting support'. Well, we can dream, anyway. That's one of the goal of jpackage project, providing a coherent set of java software packages for RPM enabled systems : http://www.jpackage.org -- To unsubscribe, e-mail: mailto:[EMAIL PROTECTED] For additional commands, e-mail: mailto:[EMAIL PROTECTED]
Re: The Complete Server Platform?
Leo, Jahia - goal:provide portal solution built on J2EE components. notes: Paul, seems like a shrink-wrap commercial solution to me??? It is. I was just pointing to it as some of the described plans/proposal were sounded like what it had done. Enhydra --- license: Enhydra Public License (Mozilla-style) No so, forked into proprietary I think. Similar story to http://www.simpledb.org/ I think. EAS --- goal:provide BSD-style license J2EE server. offers: Servlets/JSP, EJB, Avalon license: BSD-Style url: http://eas.betaversion.org status: alpha note:been dead a while now as backing company went bankrupt. OPEN ENTERPRISE DISTRIBUTION I kinda have a problem with a project that pulls toegther other projects (as it's aim). I would hope that we will one day be able to download blocks and SARs for server applications and just drop them into an apps/ dir of Phoenix. Maye my issue is that I am unsure whether this is a project hoping to make RFC complaint servercomps in one 'product' or whether it is something else. Candidate Components For Inclusion -- Jakarta: Avalon Struts Turbine Velocity James Jetspeed Slide Tomcat Some of there are dependant on the Servlet API and thus are WebApps or tools used to build webapps. Therefore they are no so big an issue as the contract for containment of a webapp is quite well defined (WAR file). Resin, Orion, Tomcat(s) Jetty, Jo can all mount them. Apache XML: Cocoon SOAP SourceForge: Enterprise Object Broker Just to remind everyone this is a transparent bean publisher that some people might want to use instead of J2EE app servers. We have three demos that work now. snip/ This e-mail is now officially more than long enough. Then trim out the previous quoted stuff dude ;-) -- To unsubscribe, e-mail: mailto:[EMAIL PROTECTED] For additional commands, e-mail: mailto:[EMAIL PROTECTED]
Re: The Complete Server Platform?
Moving slightly retrograde on the proposal (in case we missed something ...) I think my suggestion for 'alternative to J2EE' probably muddies the waters. There are a lot of candidates for inclusion, and it would be a horrible mail-fest even to think about choosing them. One focus must be to make selecting and evaluation a tool a relatively lightweight task for the intending user - because if it's part of this platform, (s)he will *know* it delivers. GJT comes to mind as something to be added to Leo's list of similar efforts, and I expect the list would end up quite long. Perhaps one thing that could be sensibly done is to strengthen the packaging and market visibility of the things in the Jakarta family ? A great deal of pre-selection has already been done, and there is already a project name - Jakarta. 'Jakarta Development Kit' might not be the best proposal, but there again ... There needn't be any intention on 'family' grounds to exclude any other toolset that was seen to be useful, but I can't quite get my head round the difficulties of choosing candidates or the weakness of too much dilution. Again, these are things that Jakarta has inherently worked through. Does this make the proposal any more practical ? Are there serious areas which Jakarta is missing ? Perhaps one of the more useful things we might be able to add are design papers un-biassed by the issues of market orthodoxies - even to the extent of pointing out areas where the Jakarta technologies aren't the best. Or is that getting too altruistic ?-) - Tim - Original Message - From: Leo Simons [EMAIL PROTECTED] To: Jakarta General List [EMAIL PROTECTED] Sent: 25 February 2002 11:00 Subject: RE: The Complete Server Platform? 'kay. Summary: (everyone, please correct and add to?) --- SIMILAR EFFORTS --- Here's a list of similar efforts that I know of... (requirments: 1) some kind of application platform 2) 100% java 3) open source) JBoss Jahia Enhydra EAS OPEN ENTERPRISE DISTRIBUTION Candidate Components For Inclusion -- Jakarta: Apache XML: SourceForge: Exolab: (...) Clearly, there are loads. We need some criteria. -- To unsubscribe, e-mail: mailto:[EMAIL PROTECTED] For additional commands, e-mail: mailto:[EMAIL PROTECTED]
Re: Re: The Complete Server Platform?
I think this would be less contentious then you think. Basically if you add the *oh the power of those who do* principal then you'll probably get some list chatter but just say are you volunteering and they'll nearly immediately shutup. If you get two volunteers in the same area then its quite simple: you have an alternative which will make for a better distribution anyhow! (You'll have the opportunity to develop better interfaces and coupling etc in the distribution or platform)... On Mon, 25 Feb 2002 15:32:12 - Tim Hyde [EMAIL PROTECTED] wrote. Moving slightly retrograde on the proposal (in case we missed something ...) I think my suggestion for 'alternative to J2EE' probably muddies the waters. There are a lot of candidates for inclusion, and it would be a horrible mail-fest even to think about choosing them. One focus must be to make selecting and evaluation a tool a relatively lightweight task for the intending user - because if it's part of this platform, (s)he will *know* it delivers. GJT comes to mind as something to be added to Leo's list of similar efforts, and I expect the list would end up quite long. Perhaps one thing that could be sensibly done is to strengthen the packaging and market visibility of the things in the Jakarta family ? A great deal of pre-selection has already been done, and there is already a project name - Jakarta. 'Jakarta Development Kit' might not be the best proposal, but there again ... There needn't be any intention on 'family' grounds to exclude any other toolset that was seen to be useful, but I can't quite get my head round the difficulties of choosing candidates or the weakness of too much dilution. Again, these are things that Jakarta has inherently worked through. Does this make the proposal any more practical ? Are there serious areas which Jakarta is missing ? Perhaps one of the more useful things we might be able to add are design papers un-biassed by the issues of market orthodoxies - even to the extent of pointing out areas where the Jakarta technologies aren't the best. Or is that getting too altruistic ?-) - Tim - Original Message - From: Leo Simons [EMAIL PROTECTED] To: Jakarta General List [EMAIL PROTECTED] Sent: 25 February 2002 11:00 Subject: RE: The Complete Server Platform? 'kay. Summary: (everyone, please correct and add to?) --- SIMILAR EFFORTS --- Here's a list of similar efforts that I know of... (requirments: 1) some kind of application platform 2) 100ava 3) open source) JBoss Jahia Enhydra EAS OPEN ENTERPRISE DISTRIBUTION Candidate Components For Inclusion -- Jakarta: Apache XML: SourceForge: Exolab: (...) Clearly, there are loads. We need some criteria. -- To unsubscribe, e-mail: mailto:[EMAIL PROTECTED] For additional commands, e-mail: mailto:[EMAIL PROTECTED] -- To unsubscribe, e-mail: mailto:[EMAIL PROTECTED] For additional commands, e-mail: mailto:[EMAIL PROTECTED]
Re: The Complete Server Platform?
Ainsi parlait GOMEZ Henri : One small extra: if a RedHat style toolkit distribution were available, the number of independent consultants who could offer their support services would exceed the number available to BEA, for example, eliminating that argument that 'I buy where I can depend on getting support'. Well, we can dream, anyway. That's one of the goal of jpackage project, providing a coherent set of java software packages for RPM enabled systems : http://www.jpackage.org And Mandrake is planning inclusion of several of these packages into next release, now that the biggest work (harmonisation and cross-dependencies computing) is done. -- Guillaume Rousse [EMAIL PROTECTED] GPG key http://lis.snv.jussieu.fr/~rousse/gpgkey.html -- To unsubscribe, e-mail: mailto:[EMAIL PROTECTED] For additional commands, e-mail: mailto:[EMAIL PROTECTED]
Re: The Complete Server Platform?
acoliver wrote: I think this would be less contentious then you think. Basically if you add the *oh the power of those who do* principal then you'll probably get some list chatter but just say are you volunteering and they'll nearly immediately shutup. +1 If you get two volunteers in the same area then its quite simple: you have an alternative which will make for a better distribution anyhow! (You'll have the opportunity to develop better interfaces and coupling etc in the distribution or platform)... -- ~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~- - Andrei (a.k.a. Andrus) Adamchik http://objectstyle.org email: andrus at objectstyle dot org -- To unsubscribe, e-mail: mailto:[EMAIL PROTECTED] For additional commands, e-mail: mailto:[EMAIL PROTECTED]
Re: The Complete Server Platform?
Leo Simons wrote: 2) a statement of intent in important places on the website. I'm guessing that putting we would like to see tomcat integrate with avalon on the projects' respective websites would mean that such will happen sooner. My concern would be that this promotes a We are Borg attitude. Why would we like to see Avalon integrate with Tomcat? Why not Jetty or Resin? If our committers are choosing to use Tomcat because it meets their real-life needs, then why would we have to tell them that. Won't they just do it because they need it? IMHO, committers should decide what is best for their product first. If we all do that, and we all create best-of-breed products, then interoperability will be a natural occurance. If it is not a natural occurance, then we are mixing politics with technology ... We would simply be replacing the Dark Lords with a Dark Lady. AFAIK, we don't tell Jakarta committers to use Ant. They choose Ant because it is a great tool. The same should go for every Jakarta or ASF product. Meanwhile, I do think documenting how J2SE (Standard Edition) technologies can provide an end-to-end solution is a great idea. For example, jGuru is running on Resin, Lucene, James, and JSPs. Scales great, and not an EJB to be found. So, should we (meaning I) write that up as a case study and post it on the site? Or, pass because they are not using Tomcat? AFAIK, Apache is part of the JS2E group, rather than the J2EE group, so it seems to me that promoting J2SE solutions is a natural thing for us to do, regardless of who provides the underlying product. -- Ted Husted, Husted dot Com, Fairport NY US -- Developing Java Web Applications with Struts -- Tel: +1 585 737-3463 -- Web: http://husted.com/struts -- To unsubscribe, e-mail: mailto:[EMAIL PROTECTED] For additional commands, e-mail: mailto:[EMAIL PROTECTED]
Re: The Complete Server Platform?
On Sun, 2002-02-24 at 10:30, Ted Husted wrote: Leo Simons wrote: 2) a statement of intent in important places on the website. I'm guessing that putting we would like to see tomcat integrate with avalon on the projects' respective websites would mean that such will happen sooner. My concern would be that this promotes a We are Borg attitude. +1 That looks like the kind of statement that cannot be reached by consensus. I'm sure there are two sides to that idea.. Why would we like to see Avalon integrate with Tomcat? Why not Jetty or Resin? *shrugs* if the Avalon folks want to see it then I guess they'd contribute to Tomcat.. If our committers are choosing to use Tomcat because it meets their real-life needs, then why would we have to tell them that. Won't they just do it because they need it? IMHO, there should be a *slant* towards inter-Apache-community cross-breeding. Not Nazi enforcement... IMHO, committers should decide what is best for their product first. If we all do that, and we all create best-of-breed products, then interoperability will be a natural occurance. +1 If it is not a natural occurance, then we are mixing politics with technology ... We would simply be replacing the Dark Lords with a Dark Lady. *shrugs* Ladies are prettier AFAIK, we don't tell Jakarta committers to use Ant. They choose Ant because it is a great tool. The same should go for every Jakarta or ASF product. +1 - Ant RULEZ! Make SUCKS! Meanwhile, I do think documenting how J2SE (Standard Edition) technologies can provide an end-to-end solution is a great idea. For example, jGuru is running on Resin, Lucene, James, and JSPs. Scales great, and not an EJB to be found. +1 So, should we (meaning I) write that up as a case study and post it on the site? Or, pass because they are not using Tomcat? I think they should be using SOME jakarta technologies otherwise it kinda doesn't look like the right site, but *shrugs*. AFAIK, Apache is part of the JS2E group, rather than the J2EE group, so it seems to me that promoting J2SE solutions is a natural thing for us to do, regardless of who provides the underlying product. IMHO: g/Apache/s//Jakarta/g (Httpd has little to do with J2-anything) Right, but I think providing server-side alternatives to J2EE is a good thing. Whether Jakarta should do it *shrug*...I think its a good idea... But its already happening -- Thanks Paul Hammant.. That seems like the best way to show the Jakarta Position to the JSPA.. There is nothing like doing something better to really make the point sink. Will it be better *shrugs* hope so.. It will certainly be more open. I applaud Paul et al's efforts and I'll come around as soon as my plate clears some (I mean it). Oh the amazing power of those who do. -Andy -- Ted Husted, Husted dot Com, Fairport NY US -- Developing Java Web Applications with Struts -- Tel: +1 585 737-3463 -- Web: http://husted.com/struts -- To unsubscribe, e-mail: mailto:[EMAIL PROTECTED] For additional commands, e-mail: mailto:[EMAIL PROTECTED] -- http://www.superlinksoftware.com http://jakarta.apache.org - port of Excel/Word/OLE 2 Compound Document format to java http://developer.java.sun.com/developer/bugParade/bugs/4487555.html - fix java generics! The avalanche has already started. It is too late for the pebbles to vote. -Ambassador Kosh -- To unsubscribe, e-mail: mailto:[EMAIL PROTECTED] For additional commands, e-mail: mailto:[EMAIL PROTECTED]
Re: The Complete Server Platform?
On 23 Feb 2002, Pete Chown wrote: The other thing I would like to push is gcj. It doesn't seem to be very well known. For people who haven't come across it, it is part of gcc and it is an ahead-of-time compiler for Java. It also includes a bytecode interpreter so it can deal with dynamically generated code, and a free implementation of the Java class libraries. And many jakarta projects compile and work fine ( and fast ) using gcj. I've got similar results with IBM's JDK1.3 for linux, which is one of the fastest ( when testing tomcat ), except the startup time which is much better. In addition there is an ant task to compile java to native using gcj ( it's part of j-t-c build process, the jkant package ). Costin -- To unsubscribe, e-mail: mailto:[EMAIL PROTECTED] For additional commands, e-mail: mailto:[EMAIL PROTECTED]
Re: The Complete Server Platform?
Ted Husted wrote: My concern would be that this promotes a We are Borg attitude. This is exactly why I think setting up a separate project would be a good idea. A project that would realize the concept of a platform that uses open tools (as opposed to jakarta-only tools). Though of course it will most likely be 70-80% Jakarta based. Here is a draft of proposed project description: --- Project XYZ (replace with a catchy name :-)) is dedicated to creation of development and deployment platform for Java web applications. It is based on other open technologies. Each one of such technologies specializes in a certain area of web application building, such as XML, HTML creation, database access, logging, Java development process support, etc. Project XYZ is a fusion of these technologies into a Java server solution that covers all parts of the spectrum. Also it introduces extensions (... need to be defined ...). Other project goals are: - to create a lower entry barrier for web application developers to use state of the art technology by providing a single full solution that works - to realize our vision of the best design and development practices - Project goals are NOT: - to set the only acceptable way of doing things in Java - to reduce the importance of technologies that are not in the core of XYZ project --- Here is a comparison to show why such a project is needed and why this is a good thing. Take Linux example. How many people in the world could take kernel code and GNU compiler and make a system that even boots? With the emergence of companies like RedHat, people received this missing piece - packaging. It was at least as important as kernel code itself. And RedHat is not the only way to package Linux, so nobody really thinks that RedHat stole Linux. I think here we have a similar situation - full platform solution will make life easier for many people, while no harm will be done to the foundation projects and their alternatives. -- ~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~- - Andrei (a.k.a. Andrus) Adamchik http://objectstyle.org list email: andrus-jk at objectstyle dot org personal email: andrus at objectstyle dot org -- To unsubscribe, e-mail: mailto:[EMAIL PROTECTED] For additional commands, e-mail: mailto:[EMAIL PROTECTED]
Re: The Complete Server Platform?
On Sun, 2002-02-24 at 12:42, Andrus Adamchik wrote: Ted Husted wrote: My concern would be that this promotes a We are Borg attitude. This is exactly why I think setting up a separate project would be a good idea. A project that would realize the concept of a platform that uses open tools (as opposed to jakarta-only tools). Though of course it will most likely be 70-80% Jakarta based. +1 -- I think eventually it could be brought to Jakarta if it was successful. Here is a draft of proposed project description: --- Project XYZ (replace with a catchy name :-)) is dedicated to creation of development and deployment platform for Java web applications. It is based on other open technologies. Each one of such technologies specializes in a certain area of web application building, such as XML, HTML creation, database access, logging, Java development process support, etc. Project XYZ is a fusion of these technologies into a Java server solution that covers all parts of the spectrum. Also it introduces extensions (... need to be defined ...). Other project goals are: - to create a lower entry barrier for web application developers to use state of the art technology by providing a single full solution that works - to realize our vision of the best design and development practices - +1 Project goals are NOT: - to set the only acceptable way of doing things in Java - to reduce the importance of technologies that are not in the core of XYZ project --- Here is a comparison to show why such a project is needed and why this is a good thing. Take Linux example. How many people in the world could take kernel code and GNU compiler and make a system that even boots? With the emergence of companies like RedHat, people received this missing piece - packaging. It was at least as important as kernel code itself. And RedHat is not the only way to package Linux, so nobody really thinks that RedHat stole Linux. -1 - You'll have some issues basing a project on GNU stuff. Legal issues. You have to make sure whatever it is -- is LGPL not GPL...otherwise you'll need good lawyers to prove the GPL virus clause invalid. I think here we have a similar situation - full platform solution will make life easier for many people, while no harm will be done to the foundation projects and their alternatives. +1 -Andy -- ~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~- - Andrei (a.k.a. Andrus) Adamchik http://objectstyle.org list email: andrus-jk at objectstyle dot org personal email: andrus at objectstyle dot org -- To unsubscribe, e-mail: mailto:[EMAIL PROTECTED] For additional commands, e-mail: mailto:[EMAIL PROTECTED] -- http://www.superlinksoftware.com http://jakarta.apache.org - port of Excel/Word/OLE 2 Compound Document format to java http://developer.java.sun.com/developer/bugParade/bugs/4487555.html - fix java generics! The avalanche has already started. It is too late for the pebbles to vote. -Ambassador Kosh -- To unsubscribe, e-mail: mailto:[EMAIL PROTECTED] For additional commands, e-mail: mailto:[EMAIL PROTECTED]
Re: The Complete Server Platform?
Andrew C. Oliver wrote: On Sun, 2002-02-24 at 12:42, Andrus Adamchik wrote This is exactly why I think setting up a separate project would be a good idea. A project that would realize the concept of a platform that uses open tools (as opposed to jakarta-only tools). Though of course it will most likely be 70-80% Jakarta based. +1 -- I think eventually it could be brought to Jakarta if it was successful I am ready to setup a project on SourceForge. Any name suggestions? How about Fusion? Kind of on topic, though sounds like ColdFusion. When the name is decided upon we can transfer this discussion to SourceForge, though I of course have a lot of hope that this community will be interested enough to provide input. Especially since this project is not so much about coding, but rather architecture. -- ~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~- - Andrei (a.k.a. Andrus) Adamchik http://objectstyle.org email: andrus at objectstyle dot org -- To unsubscribe, e-mail: mailto:[EMAIL PROTECTED] For additional commands, e-mail: mailto:[EMAIL PROTECTED]
Re: The Complete Server Platform?
Open Enterprise Distribution ... I'm bigger into descriptive names that mean something when they don't cause lawsuits... :-) You asked... What's in a name? -Andy On Sun, 2002-02-24 at 13:52, Andrus Adamchik wrote: Andrew C. Oliver wrote: On Sun, 2002-02-24 at 12:42, Andrus Adamchik wrote This is exactly why I think setting up a separate project would be a good idea. A project that would realize the concept of a platform that uses open tools (as opposed to jakarta-only tools). Though of course it will most likely be 70-80% Jakarta based. +1 -- I think eventually it could be brought to Jakarta if it was successful I am ready to setup a project on SourceForge. Any name suggestions? How about Fusion? Kind of on topic, though sounds like ColdFusion. When the name is decided upon we can transfer this discussion to SourceForge, though I of course have a lot of hope that this community will be interested enough to provide input. Especially since this project is not so much about coding, but rather architecture. -- ~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~- - Andrei (a.k.a. Andrus) Adamchik http://objectstyle.org email: andrus at objectstyle dot org -- To unsubscribe, e-mail: mailto:[EMAIL PROTECTED] For additional commands, e-mail: mailto:[EMAIL PROTECTED] -- http://www.superlinksoftware.com http://jakarta.apache.org - port of Excel/Word/OLE 2 Compound Document format to java http://developer.java.sun.com/developer/bugParade/bugs/4487555.html - fix java generics! The avalanche has already started. It is too late for the pebbles to vote. -Ambassador Kosh -- To unsubscribe, e-mail: mailto:[EMAIL PROTECTED] For additional commands, e-mail: mailto:[EMAIL PROTECTED]
Re: The Complete Server Platform?
How about a (semi?)-catchy name and a meaningful name-- Open Distribution for the Enterprise -- ODE Just my two cents, Bill Barnhill - Original Message - From: Andrew C. Oliver [EMAIL PROTECTED] To: Jakarta General List [EMAIL PROTECTED] Sent: Sunday, February 24, 2002 11:32 AM Subject: Re: The Complete Server Platform? Open Enterprise Distribution ... I'm bigger into descriptive names that mean something when they don't cause lawsuits... :-) You asked... What's in a name? -Andy On Sun, 2002-02-24 at 13:52, Andrus Adamchik wrote: Andrew C. Oliver wrote: On Sun, 2002-02-24 at 12:42, Andrus Adamchik wrote This is exactly why I think setting up a separate project would be a good idea. A project that would realize the concept of a platform that uses open tools (as opposed to jakarta-only tools). Though of course it will most likely be 70-80% Jakarta based. +1 -- I think eventually it could be brought to Jakarta if it was successful I am ready to setup a project on SourceForge. Any name suggestions? How about Fusion? Kind of on topic, though sounds like ColdFusion. When the name is decided upon we can transfer this discussion to SourceForge, though I of course have a lot of hope that this community will be interested enough to provide input. Especially since this project is not so much about coding, but rather architecture. -- ~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~- - Andrei (a.k.a. Andrus) Adamchik http://objectstyle.org email: andrus at objectstyle dot org -- To unsubscribe, e-mail: mailto:[EMAIL PROTECTED] For additional commands, e-mail: mailto:[EMAIL PROTECTED] -- http://www.superlinksoftware.com http://jakarta.apache.org - port of Excel/Word/OLE 2 Compound Document format to java http://developer.java.sun.com/developer/bugParade/bugs/4487555.html - fix java generics! The avalanche has already started. It is too late for the pebbles to vote. -Ambassador Kosh -- To unsubscribe, e-mail: mailto:[EMAIL PROTECTED] For additional commands, e-mail: mailto:[EMAIL PROTECTED] -- To unsubscribe, e-mail: mailto:[EMAIL PROTECTED] For additional commands, e-mail: mailto:[EMAIL PROTECTED]
RE: The Complete Server Platform?
QED -- Quality Enterprise Distribution (or Quite Easy Dummy as my old Philosophy 101 prof used to say). Marc Saegesser -Original Message- From: Andrew C. Oliver [mailto:[EMAIL PROTECTED]] Sent: Sunday, February 24, 2002 1:32 PM To: Jakarta General List Subject: Re: The Complete Server Platform? Open Enterprise Distribution ... I'm bigger into descriptive names that mean something when they don't cause lawsuits... :-) You asked... What's in a name? -Andy On Sun, 2002-02-24 at 13:52, Andrus Adamchik wrote: Andrew C. Oliver wrote: On Sun, 2002-02-24 at 12:42, Andrus Adamchik wrote This is exactly why I think setting up a separate project would be a good idea. A project that would realize the concept of a platform that uses open tools (as opposed to jakarta-only tools). Though of course it will most likely be 70-80% Jakarta based. +1 -- I think eventually it could be brought to Jakarta if it was successful I am ready to setup a project on SourceForge. Any name suggestions? How about Fusion? Kind of on topic, though sounds like ColdFusion. When the name is decided upon we can transfer this discussion to SourceForge, though I of course have a lot of hope that this community will be interested enough to provide input. Especially since this project is not so much about coding, but rather architecture. -- ~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~- - Andrei (a.k.a. Andrus) Adamchik http://objectstyle.org email: andrus at objectstyle dot org -- To unsubscribe, e-mail: mailto:[EMAIL PROTECTED] For additional commands, e-mail: mailto:[EMAIL PROTECTED] -- http://www.superlinksoftware.com http://jakarta.apache.org - port of Excel/Word/OLE 2 Compound Document format to java http://developer.java.sun.com/developer/bugParade/bugs/4487555.html - fix java generics! The avalanche has already started. It is too late for the pebbles to vote. -Ambassador Kosh -- To unsubscribe, e-mail: mailto:[EMAIL PROTECTED] For additional commands, e-mail: mailto:[EMAIL PROTECTED] -- To unsubscribe, e-mail: mailto:[EMAIL PROTECTED] For additional commands, e-mail: mailto:[EMAIL PROTECTED]
Re: The Complete Server Platform?
Meaningful names are OK with me too :-) . Suggested versions revolve around Enterprise Distribution idea. I personally like Open Enterprise Distribution, with OED acronym used interchangebly. I will register SF project under this name and let everybody know when the registration goes through. Bill Barnhill wrote: How about a (semi?)-catchy name and a meaningful name-- Open Distribution for the Enterprise -- ODE Just my two cents, Bill Barnhill - Original Message - From: Andrew C. Oliver [EMAIL PROTECTED] To: Jakarta General List [EMAIL PROTECTED] Sent: Sunday, February 24, 2002 11:32 AM Subject: Re: The Complete Server Platform? Open Enterprise Distribution ... I'm bigger into descriptive names that mean something when they don't cause lawsuits... :-) You asked... What's in a name? -Andy -- ~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~- - Andrei (a.k.a. Andrus) Adamchik http://objectstyle.org list email: andrus-jk at objectstyle dot org personal email: andrus at objectstyle dot org -- To unsubscribe, e-mail: mailto:[EMAIL PROTECTED] For additional commands, e-mail: mailto:[EMAIL PROTECTED]
Re: The Complete Server Platform?
Bill, Andrew, Andrus, Marc, How about a (semi?)-catchy name and a meaningful name-- Open Distribution for the Enterprise -- ODE Recap: This is a webapp project you are talking of? Or something that serves multiple server components (from Apache and others). The below was a project I very briefly worked on with many others to produce a server that presented JBoss, Tomcat, Hypersonic, Cocoon, Turbine, JetSpeed all in one VM. It was callled OpenJODA then, now with a new codebase and renamed Jahia: http://www.xo3.com:8080/ Regards, - Paul H -- To unsubscribe, e-mail: mailto:[EMAIL PROTECTED] For additional commands, e-mail: mailto:[EMAIL PROTECTED]
Re: The Complete Server Platform?
Andrus, I'm 100% behind the idea of the complete platform, but I'm worried that your proposal talks about 'Web Applications'. I believe that what's needed is an alternative to the very idea that J2EE (or even J2SE) is *the* definitive collection of java libraries, and that the project should offering a number of sensible alternatives for use in any architecture. Database access, Logging, and Development Process are three things that you've specifically mentioned that aren't particularly Web or Server oriented. Web Applications, or Server Applications, are more part of today's 'fashion' than inherent categories of how you make a computing solution, and we can expect things to move on during the lifetime of Java. Well, we can hope, anyway. :-) So, if possible, why not talk about a 'development and deployment platform for Java applications' - and then start off by assembling both the underlying 'component' toolsets and a number of combination-examples, such as the jGuru one Ted mentioned, and whatever else might emerge during the project as perhaps 'miniature live examples'. Naturally, server applications are the primary interest point initially, but it would be nice to think that the collection of tools being provided for distribution would be offered as having wide applicability. In particular, I believe that if a thing like this is available *and gets marketed* (in the Red Hat sense) properly, we could start to see the weakening of the Dilbert idea that only vendor-supplied products are 'serious' tools. This *marketing* focus is the very thing I had settled on as being the logical conclusion of the recent threads (J2EE considered harmful, EJB=Bad, etc). A way to bring the marketplace to see that there are better alternatives than the Dark Lords. Hence the marketing side (meaning actual activity to spread the word and work in PR mode with the media) needs to be a vital part of this project, needing volunteers of a different sort than technicians. But assembling the distribution first is very important, and I'm with you on this. One small extra: if a RedHat style toolkit distribution were available, the number of independent consultants who could offer their support services would exceed the number available to BEA, for example, eliminating that argument that 'I buy where I can depend on getting support'. Well, we can dream, anyway. I had been considering a project along these lines, and had thought of the name 'Tonic', both because it might revive a sickening architecture and because the Tonic (in musical terms) is where you want to go after the Dominant :-) But, if OED is the same thing, yes, what's in a name ? But, think marketing eventually !! - Tim - Original Message - From: Andrus Adamchik [EMAIL PROTECTED] To: Jakarta General List [EMAIL PROTECTED] Sent: 24 February 2002 21:38 Subject: Re: The Complete Server Platform? Meaningful names are OK with me too :-) . Suggested versions revolve around Enterprise Distribution idea. I personally like Open Enterprise Distribution, with OED acronym used interchangebly. I will register SF project under this name and let everybody know when the registration goes through. Bill Barnhill wrote: How about a (semi?)-catchy name and a meaningful name-- Open Distribution for the Enterprise -- ODE Just my two cents, Bill Barnhill - Original Message - From: Andrew C. Oliver [EMAIL PROTECTED] To: Jakarta General List [EMAIL PROTECTED] Sent: Sunday, February 24, 2002 11:32 AM Subject: Re: The Complete Server Platform? Open Enterprise Distribution ... I'm bigger into descriptive names that mean something when they don't cause lawsuits... :-) You asked... What's in a name? -Andy -- ~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~- - Andrei (a.k.a. Andrus) Adamchik http://objectstyle.org list email: andrus-jk at objectstyle dot org personal email: andrus at objectstyle dot org -- To unsubscribe, e-mail: mailto:[EMAIL PROTECTED] For additional commands, e-mail: mailto:[EMAIL PROTECTED] -- To unsubscribe, e-mail: mailto:[EMAIL PROTECTED] For additional commands, e-mail: mailto:[EMAIL PROTECTED]
Re: The Complete Server Platform?
Tim Hyde wrote: Andrus, I'm 100% behind the idea of the complete platform, but I'm worried that your proposal talks about 'Web Applications'. I believe that what's needed is an alternative to the very idea that J2EE (or even J2SE) is *the* definitive collection of java libraries, and that the project should offering a number of sensible alternatives for use in any architecture. We still will depend on certain commercial JVM's (Sun or IBM), right? Database access, Logging, and Development Process are three things that you've specifically mentioned that aren't particularly Web or Server oriented. Web Applications, or Server Applications, are more part of today's 'fashion' than inherent categories of how you make a computing solution, and we can expect things to move on during the lifetime of Java. Well, we can hope, anyway. :-) You are right. I was mentioning Web applications just cause I wanted to limit initial scope to something sane. And I guess because I am myself is a better expert in this area then in any other. This would've helped to concentrate on a certain solution-based approach from the beginning. But I agree we can widen the scope as long as we can outline the problems being solved. So, if possible, why not talk about a 'development and deployment platform for Java applications' - and then start off by assembling both the underlying 'component' toolsets and a number of combination-examples, such as the jGuru one Ted mentioned, and whatever else might emerge during the project as perhaps 'miniature live examples'. +1, like I said above, I am for it if we define use cases we are going after. Naturally, server applications are the primary interest point initially, but it would be nice to think that the collection of tools being provided for distribution would be offered as having wide applicability. In particular, I believe that if a thing like this is available *and gets marketed* (in the Red Hat sense) properly, we could start to see the weakening of the Dilbert idea that only vendor-supplied products are 'serious' tools. This *marketing* focus is the very thing I had settled on as being the logical conclusion of the recent threads (J2EE considered harmful, EJB=Bad, etc). A way to bring the marketplace to see that there are better alternatives than the Dark Lords. Hence the marketing side (meaning actual activity to spread the word and work in PR mode with the media) needs to be a vital part of this project, needing volunteers of a different sort than technicians. But assembling the distribution first is very important, and I'm with you on this. yes, this should be the starting point One small extra: if a RedHat style toolkit distribution were available, the number of independent consultants who could offer their support services would exceed the number available to BEA, for example, eliminating that argument that 'I buy where I can depend on getting support'. Well, we can dream, anyway. +1. I am an independent consultant myself, and I would stick with a technology that - allows me to concentrate on customer requirements rather then repetitive coding tasks, - offers strong design direction, - implemented most of the standard tasks already. The only thing that would prevent me from using such technology is that customer's CIO has never read about it in JDJ. I had been considering a project along these lines, and had thought of the name 'Tonic', both because it might revive a sickening architecture and because the Tonic (in musical terms) is where you want to go after the Dominant :-) But, if OED is the same thing, yes, what's in a name ? But, think marketing eventually !! I like the name Tonic, but I think Open Enterprise Distribution may actually serve that same marketing goal better. It does sound ..umm.. serious or something :-) . -- ~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~- - Andrei (a.k.a. Andrus) Adamchik http://objectstyle.org list email: andrus-jk at objectstyle dot org personal email: andrus at objectstyle dot org -- To unsubscribe, e-mail: mailto:[EMAIL PROTECTED] For additional commands, e-mail: mailto:[EMAIL PROTECTED]
Re: The Complete Server Platform?
I was thinking about another way of pushing Jakarta, partly in the context of the issue with Sun. If there was an open process for standard setting, it could make Sun's closed process less important. The IETF does well at being open, but I don't think they would get involved in something like this. Things like the RFC Editor function are already overstretched just doing network standardisation. However, something along similar organisational lines could work. The other thing I would like to push is gcj. It doesn't seem to be very well known. For people who haven't come across it, it is part of gcc and it is an ahead-of-time compiler for Java. It also includes a bytecode interpreter so it can deal with dynamically generated code, and a free implementation of the Java class libraries. To some extent a project like this is always trying to keep up with Sun, but it isn't doing badly. There is already some code (in CVS) to bring it into compliance with some aspects of JDK 1.4, for example. It will also compile many of the Jakarta products. There is a list of some current ports at: http://sources.redhat.com/rhug/ A community-based standards process, together with gcj would provide independence from Sun. It is tempting at the moment to embrace .NET, but that would just tempt Microsoft to pull the same tricks in a few years' time. Microsoft is not in this out of altruism any more than Sun is. The other point is that Sun will only change if they are given a reason to. They are a profit-making company and so of course they will act in their own interests; that is how business works. The Java Apache projects have been one of the major reasons for Java's acceptance. Apache distancing itself from Sun would be a major reason for them to be more accommodating. (Admittedly Apache doing things with .NET would be an incentive for Sun too, but my first point still holds.) Right now, Java has a major threat from .NET, and Sun really shouldn't be fighting with its allies... -- Pete -- To unsubscribe, e-mail: mailto:[EMAIL PROTECTED] For additional commands, e-mail: mailto:[EMAIL PROTECTED]
Re: The Complete Server Platform?
On Sat, 2002-02-23 at 11:01, Andrus Adamchik wrote: It is interesting that I made a similar proposal (or rather described the same idea) just yesterday on an unrelated mailing list, with the normal excuse of being too busy right now to start working in this direction ( http://groups.yahoo.com/group/webobjects/message/45867 ). I guess the idea is just flying in the air :-) I've been saying it for awhile. I'm too busy with POI, Cocoon and Lucene. (Having to learn Avalon) I'll get to it eventually I'm sure if noone else does... Of course we may all be C# programmers by then haha... Why It Doesn't Happen ...is also pretty straightforward. Individual developers work on the various Jakarta projects because they have an immediate, specific need for the project they're working on. So that's what they do. There isn't enough people around devoting energy to inter-project communication to get an integration project underway true, even though I am not a Jakarta committer, I am in a similar situation with my own project. I don't think you need to be. What To Do 1) as proposed before, a separate (from general) mailinglist dedicated to general discussion. Sharing thoughts should develop into sharing code every so often. agreed. Thats all I needone more mail list... *phew* 2) a statement of intent in important places on the website. I'm guessing that putting we would like to see tomcat integrate with avalon on the projects' respective websites would mean that such will happen sooner. Maybe this doesn't even have to be supported by individual Jakarta projects. Rather there can be a separate project with its own participants. Setting up a mailing list would help to keep design going. And after (if) a viable design/vision results from it, a project can be setup in CVS to start this big scale jakarta integration and writing any missing extensions. Thats what I think! I think it should start on sourceforge or some similar place with its own mail lists, cvs, etc. I think it should have *sub* distributions as well more specialized to different areas of the *enterprise*... etc etc. -- ~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~-~- - Andrei (a.k.a. Andrus) Adamchik http://objectstyle.org list email: andrus-jk at objectstyle dot org personal email: andrus at objectstyle dot org -- To unsubscribe, e-mail: mailto:[EMAIL PROTECTED] For additional commands, e-mail: mailto:[EMAIL PROTECTED] -- www.superlinksoftware.com www.sourceforge.net/projects/poi - port of Excel format to java http://developer.java.sun.com/developer/bugParade/bugs/4487555.html - fix java generics! The avalanche has already started. It is too late for the pebbles to vote. -Ambassador Kosh -- To unsubscribe, e-mail: mailto:[EMAIL PROTECTED] For additional commands, e-mail: mailto:[EMAIL PROTECTED]