Re: Native M3 release status
Just a quick update. The doc is in order now (thanks Andy) and the distros are ok on Linux/Mac. I'm having to rework the source distro for Windows as there were some errors in the build settings. A few more minor changes to the build script to make the extensions optional and I'll be done. Cheers, -- Pete
Re: Native M3 release status
Pete Robbins wrote: Just a quick update. The doc is in order now (thanks Andy) and the distros are ok on Linux/Mac. I'm having to rework the source distro for Windows as there were some errors in the build settings. A few more minor changes to the build script to make the extensions optional and I'll be done. Cheers, Cool. I just refreshed and it built OK on Linux, except that the PHP extension was not built at all even after doing configure --enable-all-extensions. It does not seem to be built by build.sh either but I guess it's ok... I'm assuming that you're not going to include the PHP extension in the release yet? One minor thing, extensions/ws/service/deploy.sh didn't work for me, the script looks for $APFULLDIR/modules/tuscany/module.xml which does not exist. I've tested the command line and REST samples and they worked. I can help test your release candidate on Linux when it's ready. -- Jean-Sebastien - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
Re: Native M3 release status
On 13/03/07, Jean-Sebastien Delfino [EMAIL PROTECTED] wrote: Pete Robbins wrote: Just a quick update. The doc is in order now (thanks Andy) and the distros are ok on Linux/Mac. I'm having to rework the source distro for Windows as there were some errors in the build settings. A few more minor changes to the build script to make the extensions optional and I'll be done. Cheers, Cool. I just refreshed and it built OK on Linux, except that the PHP extension was not built at all even after doing configure --enable-all-extensions. It does not seem to be built by build.sh either but I guess it's ok... I'm assuming that you're not going to include the PHP extension in the release yet? One minor thing, extensions/ws/service/deploy.sh didn't work for me, the script looks for $APFULLDIR/modules/tuscany/module.xml which does not exist. I've tested the command line and REST samples and they worked. I can help test your release candidate on Linux when it's ready. -- Jean-Sebastien PHP is not included in the main build anymore. To build it you go to sca/runtime/extensions/php and run build.sh. It has it's own autoconf/make so this does a configure/make/make install triple. It will get deployed into wherever $TUSCANY_SCACPP is set to. I'll look in to the ws deploy problem Thanks for your help. -- Pete
Re: Native M3 release status
On 13/03/07, Jean-Sebastien Delfino [EMAIL PROTECTED] wrote: Pete Robbins wrote: Just a quick update. The doc is in order now (thanks Andy) and the distros are ok on Linux/Mac. I'm having to rework the source distro for Windows as there were some errors in the build settings. A few more minor changes to the build script to make the extensions optional and I'll be done. Cheers, Cool. I just refreshed and it built OK on Linux, except that the PHP extension was not built at all even after doing configure --enable-all-extensions. It does not seem to be built by build.sh either but I guess it's ok... I'm assuming that you're not going to include the PHP extension in the release yet? One minor thing, extensions/ws/service/deploy.sh didn't work for me, the script looks for $APFULLDIR/modules/tuscany/module.xml which does not exist. This looks like I've made a typo in the Makefile.am. I have module_DATA instead of modules_DATA. I've tested the command line and REST samples and they worked. I can help test your release candidate on Linux when it's ready. -- Jean-Sebastien - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED] -- Pete
Re: Native M3 release status
[snip] Andrew Borley wrote: On 3/2/07, Pete Robbins [EMAIL PROTECTED] wrote: I can now build the PHP extension and a distro of it's source/binaries separate from the rest of the release (at leat on Linux for now). I have some other questions on how we should package the release. 1) Should we produce a source and binary download for core and each extension? This would produce many download files but would improve the modularity and flexibility of future releases. e.g. if the Ruby extension gets a major update there is no need to package, release and test the other extensions. I think this is the second time around this loop! I'm happy to go with the separate artifacts plan for the reasons you mention in 1) above, albeit with the reduced simplicity that you get with a single download package. Perhaps in the future an installer system could be used that allows users to get the kernel the extensions they require. 2) If we do 1) where should the samples go? I think the samples should belong to the extension that they are demonstrating. This means the language samples xxxCalculator would be packaged with their extension. THe REST samples would be in the REST extension (though they would pre-req a language binding e.g. Python). etc.. An alternative would be to package the samples separately. My feeling is that the samples should come with the appropriate extension. I'd propose: tuscany_sca_cpp - CppCalculator tuscany_sca_ruby - RubyCalculator tuscany_sca_python - PythonCalculator tuscany_sca_ws - CppBigBank, RubyBigBank, PythonWeatherForecast tuscany_sca_binding - HTTPDBigBank tuscany_sca_rest - RestCalculator, RestCustomer, RestYahoo, AlertAggregator Does that make sense? The WS, REST and SCA binding samples all require a language extension, but users will need at least one language extension anyway if they want to do any work with Tuscany! SCA is mainly about assembly, so most useful samples will assemble components with dependencies on different extensions (WS, REST, different scripting languages) and it's going to be difficult to package them with individual extensions, or if we really want to do that we'll have to cut the samples in small chunks and IMO they won't be very interesting anymore. 3) Does anyone ever download the Linux binary release? In my experience the download source/build/install model is used for the vast majority of Linux projects. We only produce a binary for a single Linux anyway so unless you are using RHEL3 you need to go via the source. It may make sense to have a Mac binary download but again this would be for Mac OS X Intel so of no use o the PPC Macs. This is a very valid point! I think we can drop the non-windows binaries. +1 I would like to implement 1) and have a 3 downloads per extension: Linux/Mac source, Windows source, Windows binary. Samples would be included with the relevant extensions. The extensions would be: tuscany_sca - the core tuscany_sca_cpp - C++ language binding tuscany_sca_ruby - Ruby language binding tuscany_sca_python - Python language binding tuscany_sca_ws - Axis2c webservices binding tuscany_sca_binding - sca binding (based on ws binding) tuscany_sca_rest - rest binding 3 download artifacts for each. +1 I think it makes it nice and obvious what technologies are supported. If you do that, can we keep a single download as well? I'm concerned about the complexity that we are introducing here by releasing many more small packages and asking the user to put the puzzle together. I like the user experience I get with PHP, Python, Ruby, Apache Httpd or Spring for example. One download, unzip, build, run... - on Linux, one source distribution, then the configure tool can be used to build a subset or even better it only builds what can be built with the dependencies available in your environment - on Windows, one binary distribution, function gets activated or not depending on the presence of the required dependencies (for example the PHP support gets activated if you have a PHP runtime available) Cheers Andy In regards to 3 - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED] -- Jean-Sebastien - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
Re: Native M3 release status
On 06/03/07, Jean-Sebastien Delfino [EMAIL PROTECTED] wrote: [snip] Andrew Borley wrote: On 3/2/07, Pete Robbins [EMAIL PROTECTED] wrote: I can now build the PHP extension and a distro of it's source/binaries separate from the rest of the release (at leat on Linux for now). I have some other questions on how we should package the release. 1) Should we produce a source and binary download for core and each extension? This would produce many download files but would improve the modularity and flexibility of future releases. e.g. if the Ruby extension gets a major update there is no need to package, release and test the other extensions. I think this is the second time around this loop! I'm happy to go with the separate artifacts plan for the reasons you mention in 1) above, albeit with the reduced simplicity that you get with a single download package. Perhaps in the future an installer system could be used that allows users to get the kernel the extensions they require. 2) If we do 1) where should the samples go? I think the samples should belong to the extension that they are demonstrating. This means the language samples xxxCalculator would be packaged with their extension. THe REST samples would be in the REST extension (though they would pre-req a language binding e.g. Python). etc.. An alternative would be to package the samples separately. My feeling is that the samples should come with the appropriate extension. I'd propose: tuscany_sca_cpp - CppCalculator tuscany_sca_ruby - RubyCalculator tuscany_sca_python - PythonCalculator tuscany_sca_ws - CppBigBank, RubyBigBank, PythonWeatherForecast tuscany_sca_binding - HTTPDBigBank tuscany_sca_rest - RestCalculator, RestCustomer, RestYahoo, AlertAggregator Does that make sense? The WS, REST and SCA binding samples all require a language extension, but users will need at least one language extension anyway if they want to do any work with Tuscany! SCA is mainly about assembly, so most useful samples will assemble components with dependencies on different extensions (WS, REST, different scripting languages) and it's going to be difficult to package them with individual extensions, or if we really want to do that we'll have to cut the samples in small chunks and IMO they won't be very interesting anymore. 3) Does anyone ever download the Linux binary release? In my experience the download source/build/install model is used for the vast majority of Linux projects. We only produce a binary for a single Linux anyway so unless you are using RHEL3 you need to go via the source. It may make sense to have a Mac binary download but again this would be for Mac OS X Intel so of no use o the PPC Macs. This is a very valid point! I think we can drop the non-windows binaries. +1 I would like to implement 1) and have a 3 downloads per extension: Linux/Mac source, Windows source, Windows binary. Samples would be included with the relevant extensions. The extensions would be: tuscany_sca - the core tuscany_sca_cpp - C++ language binding tuscany_sca_ruby - Ruby language binding tuscany_sca_python - Python language binding tuscany_sca_ws - Axis2c webservices binding tuscany_sca_binding - sca binding (based on ws binding) tuscany_sca_rest - rest binding 3 download artifacts for each. +1 I think it makes it nice and obvious what technologies are supported. If you do that, can we keep a single download as well? I'm concerned about the complexity that we are introducing here by releasing many more small packages and asking the user to put the puzzle together. I like the user experience I get with PHP, Python, Ruby, Apache Httpd or Spring for example. One download, unzip, build, run... - on Linux, one source distribution, then the configure tool can be used to build a subset or even better it only builds what can be built with the dependencies available in your environment - on Windows, one binary distribution, function gets activated or not depending on the presence of the required dependencies (for example the PHP support gets activated if you have a PHP runtime available) Cheers Andy In regards to 3 - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED] -- Jean-Sebastien - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED] For this (eventual) release I will be leaving everything in one download. If I can get the PHP extension working then that can be included as well. If I don't get the PHP into an acceptable shape soon I will release a package that contains no PHP and that can be released at a later date. This separate PHP package could be a good example of how to create an extension. Cheers, -- Pete
Re: Native M3 release status
On 3/2/07, Pete Robbins [EMAIL PROTECTED] wrote: On 01/03/07, Andrew Borley [EMAIL PROTECTED] wrote: On 3/1/07, Jean-Sebastien Delfino [EMAIL PROTECTED] wrote: Andrew Borley wrote: On 3/1/07, Pete Robbins [EMAIL PROTECTED] wrote: I'm just about at a point where I can produce a release candidate which includes everything except the PHP extension. I'm wondering if it would be best to publish this then release the PHP extension as a separate entity. We could go the whole hog and release a core package and then separate packages for cpp, Ruby, Python, WS binding etc.. Ultimately I think this is the way to go So If I just want to develop in Ruby and use REST I can download core, Ruby and Rest extensions and not worry about the others, and more impoortantly, their dependencies. Any thoughts? +1 from me. I've had some experience of building the PHP extension and it's quite a process - you need to build PHP with the right flags, then download and build a particular branch of the PECL SCA_SDO package and then you can build the Tuscany PHP extension! (see [1]) It may be worth waiting until the AVOCET branch of the SCA_SDO package becomes the main downloadable package from the PECL site - I believe this is the plan for the next SCA_SDO release. This (I think) will remove (or at least vastly simplify) the first 2 steps in the above process. Cheers Andy [1] http://svn.apache.org/repos/asf/incubator/tuscany/cpp/sca/runtime/extensions/php/README - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED] +1 from me. If I understand correctly then the Tuscany PHP extension will work with an actual release of the PECL SCA_SDO package. Correct? Not with the current release of the PECL SCA_SDO package (1.1.2), but it will with the next release, which will be based on the AVOCET branch that the Tuscany PHP extension depends on. Cheers Andy I can now build the PHP extension and a distro of it's source/binaries separate from the rest of the release (at leat on Linux for now). I have some other questions on how we should package the release. 1) Should we produce a source and binary download for core and each extension? This would produce many download files but would improve the modularity and flexibility of future releases. e.g. if the Ruby extension gets a major update there is no need to package, release and test the other extensions. I think this is the second time around this loop! I'm happy to go with the separate artifacts plan for the reasons you mention in 1) above, albeit with the reduced simplicity that you get with a single download package. Perhaps in the future an installer system could be used that allows users to get the kernel the extensions they require. 2) If we do 1) where should the samples go? I think the samples should belong to the extension that they are demonstrating. This means the language samples xxxCalculator would be packaged with their extension. THe REST samples would be in the REST extension (though they would pre-req a language binding e.g. Python). etc.. An alternative would be to package the samples separately. My feeling is that the samples should come with the appropriate extension. I'd propose: tuscany_sca_cpp - CppCalculator tuscany_sca_ruby - RubyCalculator tuscany_sca_python - PythonCalculator tuscany_sca_ws - CppBigBank, RubyBigBank, PythonWeatherForecast tuscany_sca_binding - HTTPDBigBank tuscany_sca_rest - RestCalculator, RestCustomer, RestYahoo, AlertAggregator Does that make sense? The WS, REST and SCA binding samples all require a language extension, but users will need at least one language extension anyway if they want to do any work with Tuscany! 3) Does anyone ever download the Linux binary release? In my experience the download source/build/install model is used for the vast majority of Linux projects. We only produce a binary for a single Linux anyway so unless you are using RHEL3 you need to go via the source. It may make sense to have a Mac binary download but again this would be for Mac OS X Intel so of no use o the PPC Macs. This is a very valid point! I think we can drop the non-windows binaries. I would like to implement 1) and have a 3 downloads per extension: Linux/Mac source, Windows source, Windows binary. Samples would be included with the relevant extensions. The extensions would be: tuscany_sca - the core tuscany_sca_cpp - C++ language binding tuscany_sca_ruby - Ruby language binding tuscany_sca_python - Python language binding tuscany_sca_ws - Axis2c webservices binding tuscany_sca_binding - sca binding (based on ws binding) tuscany_sca_rest - rest binding 3 download artifacts for each. +1 I think it makes it nice and obvious what technologies are supported. Cheers Andy In regards to 3
Re: Native M3 release status
I'm just about at a point where I can produce a release candidate which includes everything except the PHP extension. I'm wondering if it would be best to publish this then release the PHP extension as a separate entity. We could go the whole hog and release a core package and then separate packages for cpp, Ruby, Python, WS binding etc.. Ultimately I think this is the way to go So If I just want to develop in Ruby and use REST I can download core, Ruby and Rest extensions and not worry about the others, and more impoortantly, their dependencies. Any thoughts? -- Pete
Re: Native M3 release status
Andrew Borley wrote: On 3/1/07, Pete Robbins [EMAIL PROTECTED] wrote: I'm just about at a point where I can produce a release candidate which includes everything except the PHP extension. I'm wondering if it would be best to publish this then release the PHP extension as a separate entity. We could go the whole hog and release a core package and then separate packages for cpp, Ruby, Python, WS binding etc.. Ultimately I think this is the way to go So If I just want to develop in Ruby and use REST I can download core, Ruby and Rest extensions and not worry about the others, and more impoortantly, their dependencies. Any thoughts? +1 from me. I've had some experience of building the PHP extension and it's quite a process - you need to build PHP with the right flags, then download and build a particular branch of the PECL SCA_SDO package and then you can build the Tuscany PHP extension! (see [1]) It may be worth waiting until the AVOCET branch of the SCA_SDO package becomes the main downloadable package from the PECL site - I believe this is the plan for the next SCA_SDO release. This (I think) will remove (or at least vastly simplify) the first 2 steps in the above process. Cheers Andy [1] http://svn.apache.org/repos/asf/incubator/tuscany/cpp/sca/runtime/extensions/php/README - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED] +1 from me. If I understand correctly then the Tuscany PHP extension will work with an actual release of the PECL SCA_SDO package. Correct? -- Jean-Sebastien - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
Re: Native M3 release status
On 3/1/07, Jean-Sebastien Delfino [EMAIL PROTECTED] wrote: Andrew Borley wrote: On 3/1/07, Pete Robbins [EMAIL PROTECTED] wrote: I'm just about at a point where I can produce a release candidate which includes everything except the PHP extension. I'm wondering if it would be best to publish this then release the PHP extension as a separate entity. We could go the whole hog and release a core package and then separate packages for cpp, Ruby, Python, WS binding etc.. Ultimately I think this is the way to go So If I just want to develop in Ruby and use REST I can download core, Ruby and Rest extensions and not worry about the others, and more impoortantly, their dependencies. Any thoughts? +1 from me. I've had some experience of building the PHP extension and it's quite a process - you need to build PHP with the right flags, then download and build a particular branch of the PECL SCA_SDO package and then you can build the Tuscany PHP extension! (see [1]) It may be worth waiting until the AVOCET branch of the SCA_SDO package becomes the main downloadable package from the PECL site - I believe this is the plan for the next SCA_SDO release. This (I think) will remove (or at least vastly simplify) the first 2 steps in the above process. Cheers Andy [1] http://svn.apache.org/repos/asf/incubator/tuscany/cpp/sca/runtime/extensions/php/README - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED] +1 from me. If I understand correctly then the Tuscany PHP extension will work with an actual release of the PECL SCA_SDO package. Correct? Not with the current release of the PECL SCA_SDO package (1.1.2), but it will with the next release, which will be based on the AVOCET branch that the Tuscany PHP extension depends on. Cheers Andy - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
Re: Native M3 release status
On 01/03/07, Andrew Borley [EMAIL PROTECTED] wrote: On 3/1/07, Jean-Sebastien Delfino [EMAIL PROTECTED] wrote: Andrew Borley wrote: On 3/1/07, Pete Robbins [EMAIL PROTECTED] wrote: I'm just about at a point where I can produce a release candidate which includes everything except the PHP extension. I'm wondering if it would be best to publish this then release the PHP extension as a separate entity. We could go the whole hog and release a core package and then separate packages for cpp, Ruby, Python, WS binding etc.. Ultimately I think this is the way to go So If I just want to develop in Ruby and use REST I can download core, Ruby and Rest extensions and not worry about the others, and more impoortantly, their dependencies. Any thoughts? +1 from me. I've had some experience of building the PHP extension and it's quite a process - you need to build PHP with the right flags, then download and build a particular branch of the PECL SCA_SDO package and then you can build the Tuscany PHP extension! (see [1]) It may be worth waiting until the AVOCET branch of the SCA_SDO package becomes the main downloadable package from the PECL site - I believe this is the plan for the next SCA_SDO release. This (I think) will remove (or at least vastly simplify) the first 2 steps in the above process. Cheers Andy [1] http://svn.apache.org/repos/asf/incubator/tuscany/cpp/sca/runtime/extensions/php/README - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED] +1 from me. If I understand correctly then the Tuscany PHP extension will work with an actual release of the PECL SCA_SDO package. Correct? Not with the current release of the PECL SCA_SDO package (1.1.2), but it will with the next release, which will be based on the AVOCET branch that the Tuscany PHP extension depends on. Cheers Andy I can now build the PHP extension and a distro of it's source/binaries separate from the rest of the release (at leat on Linux for now). I have some other questions on how we should package the release. 1) Should we produce a source and binary download for core and each extension? This would produce many download files but would improve the modularity and flexibility of future releases. e.g. if the Ruby extension gets a major update there is no need to package, release and test the other extensions. 2) If we do 1) where should the samples go? I think the samples should belong to the extension that they are demonstrating. This means the language samples xxxCalculator would be packaged with their extension. THe REST samples would be in the REST extension (though they would pre-req a language binding e.g. Python). etc.. An alternative would be to package the samples separately. 3) Does anyone ever download the Linux binary release? In my experience the download source/build/install model is used for the vast majority of Linux projects. We only produce a binary for a single Linux anyway so unless you are using RHEL3 you need to go via the source. It may make sense to have a Mac binary download but again this would be for Mac OS X Intel so of no use o the PPC Macs. I would like to implement 1) and have a 3 downloads per extension: Linux/Mac source, Windows source, Windows binary. Samples would be included with the relevant extensions. The extensions would be: tuscany_sca - the core tuscany_sca_cpp - C++ language binding tuscany_sca_ruby - Ruby language binding tuscany_sca_python - Python language binding tuscany_sca_ws - Axis2c webservices binding tuscany_sca_binding - sca binding (based on ws binding) tuscany_sca_rest - rest binding 3 download artifacts for each. Cheers, -- Pete
Re: Native M3 release status
I think we have resolved the issues with the Python samples when mixed with Ruby and/or httpd so we have: Windows/Linux : core, cpp, ruby, python, ws, rest, sca Mac: core, cpp, ruby, python, rest Tomorrow I will try and integrate the PHP extension into the distribution and hopefully get a release candidate out this week. Cheers,
Re: Native M3 release status
On 2/28/07, Pete Robbins [EMAIL PROTECTED] wrote: I think we have resolved the issues with the Python samples when mixed with Ruby and/or httpd so we have: Windows/Linux : core, cpp, ruby, python, ws, rest, sca Mac: core, cpp, ruby, python, rest Tomorrow I will try and integrate the PHP extension into the distribution and hopefully get a release candidate out this week. Cheers, Hi Pete, What was the fix? I'm wondering whether it could also be applied to fix our lingering PHP extension linux build problem. Simon
Re: Native M3 release status
On 2/28/07, Pete Robbins [EMAIL PROTECTED] wrote: On 28/02/07, Simon Laws [EMAIL PROTECTED] wrote: On 2/28/07, Pete Robbins [EMAIL PROTECTED] wrote: I think we have resolved the issues with the Python samples when mixed with Ruby and/or httpd so we have: Windows/Linux : core, cpp, ruby, python, ws, rest, sca Mac: core, cpp, ruby, python, rest Tomorrow I will try and integrate the PHP extension into the distribution and hopefully get a release candidate out this week. Cheers, Hi Pete, What was the fix? I'm wondering whether it could also be applied to fix our lingering PHP extension linux build problem. Simon It was a mismatch in the expat library that Python includes and the one that Ruby or httpd use. To fix this we use LD_PRELOAD=our python extension before invoking Ruby (or starting httpd). This pulls in the expat that Python insists on and the other components are happy using that version. Cheers, -- Pete Ok, thanks Pete. So maybe not a solution for PHP but interesting none the less. Regards Simon
Re: Native M3 release status
Pete Robbins wrote: On 22/02/07, Pete Robbins [EMAIL PROTECTED] wrote: On 22/02/07, Jean-Sebastien Delfino [EMAIL PROTECTED] wrote: Pete Robbins wrote: On 22/02/07, Andrew Borley [EMAIL PROTECTED] wrote: On 2/21/07, Pete Robbins [EMAIL PROTECTED] wrote: On 21/02/07, Andrew Borley [EMAIL PROTECTED] wrote: On 2/21/07, Pete Robbins [EMAIL PROTECTED] wrote: On 21/02/07, Andrew Borley [EMAIL PROTECTED] wrote: On 2/21/07, Pete Robbins [EMAIL PROTECTED] wrote: I now have Ruby working on Mac. I will remove the ws bindings and clients from the *Calculator samples so we have a simple sample for each language. $TUSCANY_SCACPP/extensions/ruby/lib needs to be added to the LD_LIBRARY_PATH (or PATH on windows) to run the Ruby clients. I've updated the runclient scripts. Running a Linux build against the latest code gives me an error with the new Ruby layout: make[5]: Entering directory `/home/ajborley/workspace/TuscanyCPP/sca/runtime/extensions/ruby/extension/src' g++ -fPIC -O2 -g -pipe -Wall -Wp,-D_FORTIFY_SOURCE=2 -fexceptions -fstack-protector --param=ssp-buffer-size=4 -m32 -march=i386 -mtune=generic -fasynchronous-unwind-tables -Wall -fPIC -I. -I/usr/lib/ruby/1.8/i386-linux -I/usr/lib/ruby/1.8/i386-linux -I. -c Extension.cpp Extension.cpp:28:35: error: tuscany/sca/ruby/Ruby.h: No such file or directory Extension.cpp:29:51: error: tuscany/sca/ruby/RubyCompositeContext.h: No such file or directory Did everything get checked in? Am I missing an env variable? Everything is checked in. The extconf.rb RUby script generates the makefile. It should add the lib and include paths nexessary. Are there messages before the compile like: checking for #include tuscany/sca/ruby/RubyCompositeContext.h ? Yep, the full build log for sca/runtime/extensions/ruby/extension is: Making install in extension make[4]: Entering directory `/home/ajborley/workspace/TuscanyCPP/sca/runtime/extensions/ruby/extension' cd src; ruby extconf.rb; make checking for tuscany/sca/ruby/RubyCompositeContext.h... yes checking for main() in -ltuscany_sca_ruby_lang... yes creating Makefile make[5]: Entering directory `/home/ajborley/workspace/TuscanyCPP/sca/runtime/extensions/ruby/extension/src' g++ -fPIC -O2 -g -pipe -Wall -Wp,-D_FORTIFY_SOURCE=2 -fexceptions -fstack-protector --param=ssp-buffer-size=4 -m32 -march=i386 -mtune=generic -fasynchronous-unwind-tables -Wall -fPIC -I. -I/usr/lib/ruby/1.8/i386-linux -I/usr/lib/ruby/1.8/i386-linux -I. -c Extension.cpp Extension.cpp:28:35: error: tuscany/sca/ruby/Ruby.h: No such file or directory Extension.cpp:29:51: error: tuscany/sca/ruby/RubyCompositeContext.h: No such file or directory Extension.cpp:35: error: āVALUEā does not name a type Extension.cpp: In function āvoid Init_tuscany_sca_ruby()ā: Extension.cpp:47: error: āVALUEā was not declared in this scope Extension.cpp:47: error: expected `;' before āmoduleā Extension.cpp:48: error: āmoduleā was not declared in this scope Extension.cpp:48: error: expected primary-expression before ā)ā token Extension.cpp :48: error: āANYARGSā was not declared in this scope Extension.cpp:48: error: ārb_define_module_functionā was not declared in this scope make[5]: *** [ Extension.o] Error 1 make[5]: Leaving directory `/home/ajborley/workspace/TuscanyCPP/sca/runtime/extensions/ruby/extension/src' make[4]: *** [extension_build] Error 2 make[4]: Leaving directory `/home/ajborley/workspace/TuscanyCPP/sca/runtime/extensions/ruby/extension' make[3]: *** [install-recursive] Error 1 make[3]: Leaving directory `/home/ajborley/workspace/TuscanyCPP/sca/runtime/extensions/ruby' make[2]: *** [install-recursive] Error 1 make[2]: Leaving directory `/home/ajborley/workspace/TuscanyCPP/sca/runtime/extensions' make[1]: *** [install-recursive] Error 1 make[1]: Leaving directory `/home/ajborley/workspace/TuscanyCPP/sca/runtime' make: *** [install-recursive] Error 1 So the check for tuscany/sca/ruby/RubyCompositeContext.h is there (although there isn't one for Ruby.h) and looks like it succeeds, but the build still seems to fail to find it. The checks are supposed to add the -I and -L stuff for where it finds the headers and libraries so here should be a -I../../src in the command line. This works on Mac and my RHEL3 linux. Which linux are you running? I'm on Fedora core 5 with Ruby 1.8.5 I've checked in a new version of extconf.rb which is more simple and works on Mac and RHEL. Cheers,
Re: Native M3 release status
On 21/02/07, Pete Robbins [EMAIL PROTECTED] wrote: The Rest binding builds on Mac but is crashing when the SCARuntime loads the extension which tries to register the reference binding extension. I'll investigate that next. The Rest binding is crashing because of the way TuscanyRuntime loads the extensions. We load every library we find under the extensions folder. Unfortunately one of these is tuscany_sca_mod_rest which is not an extension but does link with one. So we load this library and discover it isn't an extension, then unload it. Next up we load the tuscany_sca_rest_service extension but for some reason the load is not calling the static initializers, maybe because this library has already been loaded as a separate instance when loading mod_rest. Removing the tuscany_sca_mod_rest library from the extension path resolves the problem. This goes back to a problem discussed on an earlier thread where we agreed to change the extension loading so that only libraries in a modules folder would be loaded. The layout of an extension will look like: $TUSCANY_SCACPP/extensions myExtension/ lib/ mylibs_that_arent_extensions.so bin/ ... xsd/ any_other_folder/ ... modules/ libmyExtensionLibrary.so Only the libraries under modules will be loaded. Most likely these would be symbolic links to the library in the lib/ folder. I propose to make this change sometime today as it resolves a few problems. Cheers, -- Pete
Re: Native M3 release status
On 2/21/07, Pete Robbins [EMAIL PROTECTED] wrote: On 21/02/07, Andrew Borley [EMAIL PROTECTED] wrote: On 2/21/07, Pete Robbins [EMAIL PROTECTED] wrote: On 21/02/07, Andrew Borley [EMAIL PROTECTED] wrote: On 2/21/07, Pete Robbins [EMAIL PROTECTED] wrote: I now have Ruby working on Mac. I will remove the ws bindings and clients from the *Calculator samples so we have a simple sample for each language. $TUSCANY_SCACPP/extensions/ruby/lib needs to be added to the LD_LIBRARY_PATH (or PATH on windows) to run the Ruby clients. I've updated the runclient scripts. Running a Linux build against the latest code gives me an error with the new Ruby layout: make[5]: Entering directory `/home/ajborley/workspace/TuscanyCPP/sca/runtime/extensions/ruby/extension/src' g++ -fPIC -O2 -g -pipe -Wall -Wp,-D_FORTIFY_SOURCE=2 -fexceptions -fstack-protector --param=ssp-buffer-size=4 -m32 -march=i386 -mtune=generic -fasynchronous-unwind-tables -Wall -fPIC -I. -I/usr/lib/ruby/1.8/i386-linux -I/usr/lib/ruby/1.8/i386-linux -I. -c Extension.cpp Extension.cpp:28:35: error: tuscany/sca/ruby/Ruby.h: No such file or directory Extension.cpp:29:51: error: tuscany/sca/ruby/RubyCompositeContext.h: No such file or directory Did everything get checked in? Am I missing an env variable? Everything is checked in. The extconf.rb RUby script generates the makefile. It should add the lib and include paths nexessary. Are there messages before the compile like: checking for #include tuscany/sca/ruby/RubyCompositeContext.h ? Yep, the full build log for sca/runtime/extensions/ruby/extension is: Making install in extension make[4]: Entering directory `/home/ajborley/workspace/TuscanyCPP/sca/runtime/extensions/ruby/extension' cd src; ruby extconf.rb; make checking for tuscany/sca/ruby/RubyCompositeContext.h... yes checking for main() in -ltuscany_sca_ruby_lang... yes creating Makefile make[5]: Entering directory `/home/ajborley/workspace/TuscanyCPP/sca/runtime/extensions/ruby/extension/src' g++ -fPIC -O2 -g -pipe -Wall -Wp,-D_FORTIFY_SOURCE=2 -fexceptions -fstack-protector --param=ssp-buffer-size=4 -m32 -march=i386 -mtune=generic -fasynchronous-unwind-tables -Wall -fPIC -I. -I/usr/lib/ruby/1.8/i386-linux -I/usr/lib/ruby/1.8/i386-linux -I. -c Extension.cpp Extension.cpp:28:35: error: tuscany/sca/ruby/Ruby.h: No such file or directory Extension.cpp:29:51: error: tuscany/sca/ruby/RubyCompositeContext.h: No such file or directory Extension.cpp:35: error: âVALUEâ does not name a type Extension.cpp: In function âvoid Init_tuscany_sca_ruby()â: Extension.cpp:47: error: âVALUEâ was not declared in this scope Extension.cpp:47: error: expected `;' before âmoduleâ Extension.cpp:48: error: âmoduleâ was not declared in this scope Extension.cpp:48: error: expected primary-expression before â)â token Extension.cpp:48: error: âANYARGSâ was not declared in this scope Extension.cpp:48: error: ârb_define_module_functionâ was not declared in this scope make[5]: *** [Extension.o] Error 1 make[5]: Leaving directory `/home/ajborley/workspace/TuscanyCPP/sca/runtime/extensions/ruby/extension/src' make[4]: *** [extension_build] Error 2 make[4]: Leaving directory `/home/ajborley/workspace/TuscanyCPP/sca/runtime/extensions/ruby/extension' make[3]: *** [install-recursive] Error 1 make[3]: Leaving directory `/home/ajborley/workspace/TuscanyCPP/sca/runtime/extensions/ruby' make[2]: *** [install-recursive] Error 1 make[2]: Leaving directory `/home/ajborley/workspace/TuscanyCPP/sca/runtime/extensions' make[1]: *** [install-recursive] Error 1 make[1]: Leaving directory `/home/ajborley/workspace/TuscanyCPP/sca/runtime' make: *** [install-recursive] Error 1 So the check for tuscany/sca/ruby/RubyCompositeContext.h is there (although there isn't one for Ruby.h) and looks like it succeeds, but the build still seems to fail to find it. The checks are supposed to add the -I and -L stuff for where it finds the headers and libraries so here should be a -I../../src in the command line. This works on Mac and my RHEL3 linux. Which linux are you running? I'm on Fedora core 5 with Ruby 1.8.5 Cheers Andy - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
Re: Native M3 release status
On 2/22/07, Pete Robbins [EMAIL PROTECTED] wrote: On 21/02/07, Pete Robbins [EMAIL PROTECTED] wrote: The Rest binding builds on Mac but is crashing when the SCARuntime loads the extension which tries to register the reference binding extension. I'll investigate that next. The Rest binding is crashing because of the way TuscanyRuntime loads the extensions. We load every library we find under the extensions folder. Unfortunately one of these is tuscany_sca_mod_rest which is not an extension but does link with one. So we load this library and discover it isn't an extension, then unload it. Next up we load the tuscany_sca_rest_service extension but for some reason the load is not calling the static initializers, maybe because this library has already been loaded as a separate instance when loading mod_rest. Removing the tuscany_sca_mod_rest library from the extension path resolves the problem. This goes back to a problem discussed on an earlier thread where we agreed to change the extension loading so that only libraries in a modules folder would be loaded. The layout of an extension will look like: $TUSCANY_SCACPP/extensions myExtension/ lib/ mylibs_that_arent_extensions.so bin/ ... xsd/ any_other_folder/ ... modules/ libmyExtensionLibrary.so Only the libraries under modules will be loaded. Most likely these would be symbolic links to the library in the lib/ folder. I propose to make this change sometime today as it resolves a few problems. +1 for this. Andy - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
Re: Native M3 release status
On 22/02/07, Andrew Borley [EMAIL PROTECTED] wrote: On 2/21/07, Pete Robbins [EMAIL PROTECTED] wrote: On 21/02/07, Andrew Borley [EMAIL PROTECTED] wrote: On 2/21/07, Pete Robbins [EMAIL PROTECTED] wrote: On 21/02/07, Andrew Borley [EMAIL PROTECTED] wrote: On 2/21/07, Pete Robbins [EMAIL PROTECTED] wrote: I now have Ruby working on Mac. I will remove the ws bindings and clients from the *Calculator samples so we have a simple sample for each language. $TUSCANY_SCACPP/extensions/ruby/lib needs to be added to the LD_LIBRARY_PATH (or PATH on windows) to run the Ruby clients. I've updated the runclient scripts. Running a Linux build against the latest code gives me an error with the new Ruby layout: make[5]: Entering directory `/home/ajborley/workspace/TuscanyCPP/sca/runtime/extensions/ruby/extension/src' g++ -fPIC -O2 -g -pipe -Wall -Wp,-D_FORTIFY_SOURCE=2 -fexceptions -fstack-protector --param=ssp-buffer-size=4 -m32 -march=i386 -mtune=generic -fasynchronous-unwind-tables -Wall -fPIC -I. -I/usr/lib/ruby/1.8/i386-linux -I/usr/lib/ruby/1.8/i386-linux -I. -c Extension.cpp Extension.cpp:28:35: error: tuscany/sca/ruby/Ruby.h: No such file or directory Extension.cpp:29:51: error: tuscany/sca/ruby/RubyCompositeContext.h: No such file or directory Did everything get checked in? Am I missing an env variable? Everything is checked in. The extconf.rb RUby script generates the makefile. It should add the lib and include paths nexessary. Are there messages before the compile like: checking for #include tuscany/sca/ruby/RubyCompositeContext.h ? Yep, the full build log for sca/runtime/extensions/ruby/extension is: Making install in extension make[4]: Entering directory `/home/ajborley/workspace/TuscanyCPP/sca/runtime/extensions/ruby/extension' cd src; ruby extconf.rb; make checking for tuscany/sca/ruby/RubyCompositeContext.h... yes checking for main() in -ltuscany_sca_ruby_lang... yes creating Makefile make[5]: Entering directory `/home/ajborley/workspace/TuscanyCPP/sca/runtime/extensions/ruby/extension/src' g++ -fPIC -O2 -g -pipe -Wall -Wp,-D_FORTIFY_SOURCE=2 -fexceptions -fstack-protector --param=ssp-buffer-size=4 -m32 -march=i386 -mtune=generic -fasynchronous-unwind-tables -Wall -fPIC -I. -I/usr/lib/ruby/1.8/i386-linux -I/usr/lib/ruby/1.8/i386-linux -I. -c Extension.cpp Extension.cpp:28:35: error: tuscany/sca/ruby/Ruby.h: No such file or directory Extension.cpp:29:51: error: tuscany/sca/ruby/RubyCompositeContext.h: No such file or directory Extension.cpp:35: error: âVALUEâ does not name a type Extension.cpp: In function âvoid Init_tuscany_sca_ruby()â: Extension.cpp:47: error: âVALUEâ was not declared in this scope Extension.cpp:47: error: expected `;' before âmoduleâ Extension.cpp:48: error: âmoduleâ was not declared in this scope Extension.cpp:48: error: expected primary-expression before â)â token Extension.cpp:48: error: âANYARGSâ was not declared in this scope Extension.cpp:48: error: ârb_define_module_functionâ was not declared in this scope make[5]: *** [Extension.o] Error 1 make[5]: Leaving directory `/home/ajborley/workspace/TuscanyCPP/sca/runtime/extensions/ruby/extension/src' make[4]: *** [extension_build] Error 2 make[4]: Leaving directory `/home/ajborley/workspace/TuscanyCPP/sca/runtime/extensions/ruby/extension' make[3]: *** [install-recursive] Error 1 make[3]: Leaving directory `/home/ajborley/workspace/TuscanyCPP/sca/runtime/extensions/ruby' make[2]: *** [install-recursive] Error 1 make[2]: Leaving directory `/home/ajborley/workspace/TuscanyCPP/sca/runtime/extensions' make[1]: *** [install-recursive] Error 1 make[1]: Leaving directory `/home/ajborley/workspace/TuscanyCPP/sca/runtime' make: *** [install-recursive] Error 1 So the check for tuscany/sca/ruby/RubyCompositeContext.h is there (although there isn't one for Ruby.h) and looks like it succeeds, but the build still seems to fail to find it. The checks are supposed to add the -I and -L stuff for where it finds the headers and libraries so here should be a -I../../src in the command line. This works on Mac and my RHEL3 linux. Which linux are you running? I'm on Fedora core 5 with Ruby 1.8.5 I've checked in a new version of extconf.rb which is more simple and works on Mac and RHEL. Cheers, -- Pete
Re: Native M3 release status
Pete Robbins wrote: On 21/02/07, Pete Robbins [EMAIL PROTECTED] wrote: The Rest binding builds on Mac but is crashing when the SCARuntime loads the extension which tries to register the reference binding extension. I'll investigate that next. The Rest binding is crashing because of the way TuscanyRuntime loads the extensions. We load every library we find under the extensions folder. Unfortunately one of these is tuscany_sca_mod_rest which is not an extension but does link with one. So we load this library and discover it isn't an extension, then unload it. Next up we load the tuscany_sca_rest_service extension but for some reason the load is not calling the static initializers, maybe because this library has already been loaded as a separate instance when loading mod_rest. Removing the tuscany_sca_mod_rest library from the extension path resolves the problem. This goes back to a problem discussed on an earlier thread where we agreed to change the extension loading so that only libraries in a modules folder would be loaded. The layout of an extension will look like: $TUSCANY_SCACPP/extensions myExtension/ lib/ mylibs_that_arent_extensions.so bin/ ... xsd/ any_other_folder/ ... modules/ libmyExtensionLibrary.so Only the libraries under modules will be loaded. Most likely these would be symbolic links to the library in the lib/ folder. I propose to make this change sometime today as it resolves a few problems. Cheers, Sounds good, +1 from me. -- Jean-Sebastien - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
Re: Native M3 release status
Pete Robbins wrote: On 22/02/07, Andrew Borley [EMAIL PROTECTED] wrote: On 2/21/07, Pete Robbins [EMAIL PROTECTED] wrote: On 21/02/07, Andrew Borley [EMAIL PROTECTED] wrote: On 2/21/07, Pete Robbins [EMAIL PROTECTED] wrote: On 21/02/07, Andrew Borley [EMAIL PROTECTED] wrote: On 2/21/07, Pete Robbins [EMAIL PROTECTED] wrote: I now have Ruby working on Mac. I will remove the ws bindings and clients from the *Calculator samples so we have a simple sample for each language. $TUSCANY_SCACPP/extensions/ruby/lib needs to be added to the LD_LIBRARY_PATH (or PATH on windows) to run the Ruby clients. I've updated the runclient scripts. Running a Linux build against the latest code gives me an error with the new Ruby layout: make[5]: Entering directory `/home/ajborley/workspace/TuscanyCPP/sca/runtime/extensions/ruby/extension/src' g++ -fPIC -O2 -g -pipe -Wall -Wp,-D_FORTIFY_SOURCE=2 -fexceptions -fstack-protector --param=ssp-buffer-size=4 -m32 -march=i386 -mtune=generic -fasynchronous-unwind-tables -Wall -fPIC -I. -I/usr/lib/ruby/1.8/i386-linux -I/usr/lib/ruby/1.8/i386-linux -I. -c Extension.cpp Extension.cpp:28:35: error: tuscany/sca/ruby/Ruby.h: No such file or directory Extension.cpp:29:51: error: tuscany/sca/ruby/RubyCompositeContext.h: No such file or directory Did everything get checked in? Am I missing an env variable? Everything is checked in. The extconf.rb RUby script generates the makefile. It should add the lib and include paths nexessary. Are there messages before the compile like: checking for #include tuscany/sca/ruby/RubyCompositeContext.h ? Yep, the full build log for sca/runtime/extensions/ruby/extension is: Making install in extension make[4]: Entering directory `/home/ajborley/workspace/TuscanyCPP/sca/runtime/extensions/ruby/extension' cd src; ruby extconf.rb; make checking for tuscany/sca/ruby/RubyCompositeContext.h... yes checking for main() in -ltuscany_sca_ruby_lang... yes creating Makefile make[5]: Entering directory `/home/ajborley/workspace/TuscanyCPP/sca/runtime/extensions/ruby/extension/src' g++ -fPIC -O2 -g -pipe -Wall -Wp,-D_FORTIFY_SOURCE=2 -fexceptions -fstack-protector --param=ssp-buffer-size=4 -m32 -march=i386 -mtune=generic -fasynchronous-unwind-tables -Wall -fPIC -I. -I/usr/lib/ruby/1.8/i386-linux -I/usr/lib/ruby/1.8/i386-linux -I. -c Extension.cpp Extension.cpp:28:35: error: tuscany/sca/ruby/Ruby.h: No such file or directory Extension.cpp:29:51: error: tuscany/sca/ruby/RubyCompositeContext.h: No such file or directory Extension.cpp:35: error: âVALUEâ does not name a type Extension.cpp: In function âvoid Init_tuscany_sca_ruby()â: Extension.cpp:47: error: âVALUEâ was not declared in this scope Extension.cpp:47: error: expected `;' before âmoduleâ Extension.cpp:48: error: âmoduleâ was not declared in this scope Extension.cpp:48: error: expected primary-expression before â)â token Extension.cpp:48: error: âANYARGSâ was not declared in this scope Extension.cpp:48: error: ârb_define_module_functionâ was not declared in this scope make[5]: *** [Extension.o] Error 1 make[5]: Leaving directory `/home/ajborley/workspace/TuscanyCPP/sca/runtime/extensions/ruby/extension/src' make[4]: *** [extension_build] Error 2 make[4]: Leaving directory `/home/ajborley/workspace/TuscanyCPP/sca/runtime/extensions/ruby/extension' make[3]: *** [install-recursive] Error 1 make[3]: Leaving directory `/home/ajborley/workspace/TuscanyCPP/sca/runtime/extensions/ruby' make[2]: *** [install-recursive] Error 1 make[2]: Leaving directory `/home/ajborley/workspace/TuscanyCPP/sca/runtime/extensions' make[1]: *** [install-recursive] Error 1 make[1]: Leaving directory `/home/ajborley/workspace/TuscanyCPP/sca/runtime' make: *** [install-recursive] Error 1 So the check for tuscany/sca/ruby/RubyCompositeContext.h is there (although there isn't one for Ruby.h) and looks like it succeeds, but the build still seems to fail to find it. The checks are supposed to add the -I and -L stuff for where it finds the headers and libraries so here should be a -I../../src in the command line. This works on Mac and my RHEL3 linux. Which linux are you running? I'm on Fedora core 5 with Ruby 1.8.5 I've checked in a new version of extconf.rb which is more simple and works on Mac and RHEL. Cheers, Hi, It looks like we have 3 sets of changes going on: - packaging changes for the extensions - Ruby build changes - PHP patches? Could you guys please post a message here when these changes are in (even before you have an RC ready)? I have Ruby 1.8.5 and PHP 5.2 installed on my Linux RHEL4 and can build and try things out if it helps. Thanks, -- Jean-Sebastien - To
Re: Native M3 release status
On 22/02/07, Jean-Sebastien Delfino [EMAIL PROTECTED] wrote: Pete Robbins wrote: On 22/02/07, Andrew Borley [EMAIL PROTECTED] wrote: On 2/21/07, Pete Robbins [EMAIL PROTECTED] wrote: On 21/02/07, Andrew Borley [EMAIL PROTECTED] wrote: On 2/21/07, Pete Robbins [EMAIL PROTECTED] wrote: On 21/02/07, Andrew Borley [EMAIL PROTECTED] wrote: On 2/21/07, Pete Robbins [EMAIL PROTECTED] wrote: I now have Ruby working on Mac. I will remove the ws bindings and clients from the *Calculator samples so we have a simple sample for each language. $TUSCANY_SCACPP/extensions/ruby/lib needs to be added to the LD_LIBRARY_PATH (or PATH on windows) to run the Ruby clients. I've updated the runclient scripts. Running a Linux build against the latest code gives me an error with the new Ruby layout: make[5]: Entering directory `/home/ajborley/workspace/TuscanyCPP/sca/runtime/extensions/ruby/extension/src' g++ -fPIC -O2 -g -pipe -Wall -Wp,-D_FORTIFY_SOURCE=2 -fexceptions -fstack-protector --param=ssp-buffer-size=4 -m32 -march=i386 -mtune=generic -fasynchronous-unwind-tables -Wall -fPIC -I. -I/usr/lib/ruby/1.8/i386-linux -I/usr/lib/ruby/1.8/i386-linux -I. -c Extension.cpp Extension.cpp:28:35: error: tuscany/sca/ruby/Ruby.h: No such file or directory Extension.cpp:29:51: error: tuscany/sca/ruby/RubyCompositeContext.h: No such file or directory Did everything get checked in? Am I missing an env variable? Everything is checked in. The extconf.rb RUby script generates the makefile. It should add the lib and include paths nexessary. Are there messages before the compile like: checking for #include tuscany/sca/ruby/RubyCompositeContext.h ? Yep, the full build log for sca/runtime/extensions/ruby/extension is: Making install in extension make[4]: Entering directory `/home/ajborley/workspace/TuscanyCPP/sca/runtime/extensions/ruby/extension' cd src; ruby extconf.rb; make checking for tuscany/sca/ruby/RubyCompositeContext.h... yes checking for main() in -ltuscany_sca_ruby_lang... yes creating Makefile make[5]: Entering directory `/home/ajborley/workspace/TuscanyCPP/sca/runtime/extensions/ruby/extension/src' g++ -fPIC -O2 -g -pipe -Wall -Wp,-D_FORTIFY_SOURCE=2 -fexceptions -fstack-protector --param=ssp-buffer-size=4 -m32 -march=i386 -mtune=generic -fasynchronous-unwind-tables -Wall -fPIC -I. -I/usr/lib/ruby/1.8/i386-linux -I/usr/lib/ruby/1.8/i386-linux -I. -c Extension.cpp Extension.cpp:28:35: error: tuscany/sca/ruby/Ruby.h: No such file or directory Extension.cpp:29:51: error: tuscany/sca/ruby/RubyCompositeContext.h: No such file or directory Extension.cpp:35: error: āVALUEā does not name a type Extension.cpp: In function āvoid Init_tuscany_sca_ruby()ā: Extension.cpp:47: error: āVALUEā was not declared in this scope Extension.cpp:47: error: expected `;' before āmoduleā Extension.cpp:48: error: āmoduleā was not declared in this scope Extension.cpp:48: error: expected primary-expression before ā)ā token Extension.cpp:48: error: āANYARGSā was not declared in this scope Extension.cpp:48: error: ārb_define_module_functionā was not declared in this scope make[5]: *** [Extension.o] Error 1 make[5]: Leaving directory `/home/ajborley/workspace/TuscanyCPP/sca/runtime/extensions/ruby/extension/src' make[4]: *** [extension_build] Error 2 make[4]: Leaving directory `/home/ajborley/workspace/TuscanyCPP/sca/runtime/extensions/ruby/extension' make[3]: *** [install-recursive] Error 1 make[3]: Leaving directory `/home/ajborley/workspace/TuscanyCPP/sca/runtime/extensions/ruby' make[2]: *** [install-recursive] Error 1 make[2]: Leaving directory `/home/ajborley/workspace/TuscanyCPP/sca/runtime/extensions' make[1]: *** [install-recursive] Error 1 make[1]: Leaving directory `/home/ajborley/workspace/TuscanyCPP/sca/runtime' make: *** [install-recursive] Error 1 So the check for tuscany/sca/ruby/RubyCompositeContext.h is there (although there isn't one for Ruby.h) and looks like it succeeds, but the build still seems to fail to find it. The checks are supposed to add the -I and -L stuff for where it finds the headers and libraries so here should be a -I../../src in the command line. This works on Mac and my RHEL3 linux. Which linux are you running? I'm on Fedora core 5 with Ruby 1.8.5 I've checked in a new version of extconf.rb which is more simple and works on Mac and RHEL. Cheers, Hi, It looks like we have 3 sets of changes going on: - packaging changes for the extensions - Ruby build changes - PHP patches? Could you guys please post a message here when these changes are in (even before you have an RC ready)? I have Ruby 1.8.5 and PHP 5.2 installed on my Linux
Re: Native M3 release status
On 2/22/07, Pete Robbins [EMAIL PROTECTED] wrote: On 22/02/07, Jean-Sebastien Delfino [EMAIL PROTECTED] wrote: Pete Robbins wrote: On 22/02/07, Andrew Borley [EMAIL PROTECTED] wrote: On 2/21/07, Pete Robbins [EMAIL PROTECTED] wrote: On 21/02/07, Andrew Borley [EMAIL PROTECTED] wrote: On 2/21/07, Pete Robbins [EMAIL PROTECTED] wrote: On 21/02/07, Andrew Borley [EMAIL PROTECTED] wrote: On 2/21/07, Pete Robbins [EMAIL PROTECTED] wrote: I now have Ruby working on Mac. I will remove the ws bindings and clients from the *Calculator samples so we have a simple sample for each language. $TUSCANY_SCACPP/extensions/ruby/lib needs to be added to the LD_LIBRARY_PATH (or PATH on windows) to run the Ruby clients. I've updated the runclient scripts. Running a Linux build against the latest code gives me an error with the new Ruby layout: make[5]: Entering directory `/home/ajborley/workspace/TuscanyCPP/sca/runtime/extensions/ruby/extension/src' g++ -fPIC -O2 -g -pipe -Wall -Wp,-D_FORTIFY_SOURCE=2 -fexceptions -fstack-protector --param=ssp-buffer-size=4 -m32 -march=i386 -mtune=generic -fasynchronous-unwind-tables -Wall -fPIC -I. -I/usr/lib/ruby/1.8/i386-linux -I/usr/lib/ruby/1.8/i386-linux -I. -c Extension.cpp Extension.cpp:28:35: error: tuscany/sca/ruby/Ruby.h: No such file or directory Extension.cpp:29:51: error: tuscany/sca/ruby/RubyCompositeContext.h: No such file or directory Did everything get checked in? Am I missing an env variable? Everything is checked in. The extconf.rb RUby script generates the makefile. It should add the lib and include paths nexessary. Are there messages before the compile like: checking for #include tuscany/sca/ruby/RubyCompositeContext.h ? Yep, the full build log for sca/runtime/extensions/ruby/extension is: Making install in extension make[4]: Entering directory `/home/ajborley/workspace/TuscanyCPP/sca/runtime/extensions/ruby/extension' cd src; ruby extconf.rb; make checking for tuscany/sca/ruby/RubyCompositeContext.h... yes checking for main() in -ltuscany_sca_ruby_lang... yes creating Makefile make[5]: Entering directory `/home/ajborley/workspace/TuscanyCPP/sca/runtime/extensions/ruby/extension/src' g++ -fPIC -O2 -g -pipe -Wall -Wp,-D_FORTIFY_SOURCE=2 -fexceptions -fstack-protector --param=ssp-buffer-size=4 -m32 -march=i386 -mtune=generic -fasynchronous-unwind-tables -Wall -fPIC -I. -I/usr/lib/ruby/1.8/i386-linux -I/usr/lib/ruby/1.8/i386-linux -I. -c Extension.cpp Extension.cpp:28:35: error: tuscany/sca/ruby/Ruby.h: No such file or directory Extension.cpp:29:51: error: tuscany/sca/ruby/RubyCompositeContext.h: No such file or directory Extension.cpp:35: error: āVALUEā does not name a type Extension.cpp: In function āvoid Init_tuscany_sca_ruby()ā: Extension.cpp:47: error: āVALUEā was not declared in this scope Extension.cpp:47: error: expected `;' before āmoduleā Extension.cpp:48: error: āmoduleā was not declared in this scope Extension.cpp:48: error: expected primary-expression before ā)ā token Extension.cpp:48: error: āANYARGSā was not declared in this scope Extension.cpp:48: error: ārb_define_module_functionā was not declared in this scope make[5]: *** [Extension.o] Error 1 make[5]: Leaving directory `/home/ajborley/workspace/TuscanyCPP/sca/runtime/extensions/ruby/extension/src' make[4]: *** [extension_build] Error 2 make[4]: Leaving directory `/home/ajborley/workspace/TuscanyCPP/sca/runtime/extensions/ruby/extension' make[3]: *** [install-recursive] Error 1 make[3]: Leaving directory `/home/ajborley/workspace/TuscanyCPP/sca/runtime/extensions/ruby' make[2]: *** [install-recursive] Error 1 make[2]: Leaving directory `/home/ajborley/workspace/TuscanyCPP/sca/runtime/extensions' make[1]: *** [install-recursive] Error 1 make[1]: Leaving directory `/home/ajborley/workspace/TuscanyCPP/sca/runtime' make: *** [install-recursive] Error 1 So the check for tuscany/sca/ruby/RubyCompositeContext.h is there (although there isn't one for Ruby.h) and looks like it succeeds, but the build still seems to fail to find it. The checks are supposed to add the -I and -L stuff for where it finds the headers and libraries so here should be a -I../../src in the command line. This works on Mac and my RHEL3 linux. Which linux are you running? I'm on Fedora core 5 with Ruby 1.8.5 I've checked in a new version of extconf.rb which is more simple and works on Mac and RHEL. Cheers, Hi, It looks like we have 3 sets of changes going on: - packaging changes for the
Re: Native M3 release status
On 2/20/07, Andrew Borley [EMAIL PROTECTED] wrote: On 2/20/07, Pete Robbins [EMAIL PROTECTED] wrote: I'm currently working through building/testing the extensions and samples across Windows, Linux and Mac OS. Cpp: extension and samples look good (CppBigBank does not run on Mac as we have no ws binding extension) Ruby: Works fine on Windows and Linux (after fix I applied yesterday). Currently re-structuring the extension to work with Mac, I think this should be done by tomorrow. Python: Basic sample runs fine across all platforms. PHP: ? - I haven't built this yet REST: partially tested on Win and Linux So... there's still a fair bit to get through. Anyone care to fill in some of the blanks? Cheers, -- Pete All the WS, REST and Python samples run fine on my Windows system, but I'm having a Python issue on Linux (I think because my Python is v2.4 and it seems hard to get it updated..) Cheers Andy On further investigation, it's actually not vary hard to update to Python 2.5 (well, run 2.5 alongside 2.4). What is hard is getting the Python ElementTree package to run happily within a Ruby environment. This is the case in the RestYahoo sample where we have a Ruby local client invoking a Python component. I get an exception thrown within the ElementTree code when it tries to load the xml.parsers.expat module (which is a binary library). I think this is due to Python needing to use its newer version of the expat lib than the one that Ruby must have loaded. There doesn't seem to be a way of upgrading the expat package on my system - the 1.95 version I have is required by 400 other packages! We could change the sample to have a Python client instead of Ruby, as it is more about demonstrating the REST extension rather than Ruby talking to Python, but it looks like we have a restriction here. Anyone have any ideas? Cheers Andy - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
Re: Native M3 release status
On 21/02/07, Andrew Borley [EMAIL PROTECTED] wrote: On 2/20/07, Andrew Borley [EMAIL PROTECTED] wrote: On 2/20/07, Pete Robbins [EMAIL PROTECTED] wrote: I'm currently working through building/testing the extensions and samples across Windows, Linux and Mac OS. Cpp: extension and samples look good (CppBigBank does not run on Mac as we have no ws binding extension) Ruby: Works fine on Windows and Linux (after fix I applied yesterday). Currently re-structuring the extension to work with Mac, I think this should be done by tomorrow. Python: Basic sample runs fine across all platforms. PHP: ? - I haven't built this yet REST: partially tested on Win and Linux So... there's still a fair bit to get through. Anyone care to fill in some of the blanks? Cheers, -- Pete All the WS, REST and Python samples run fine on my Windows system, but I'm having a Python issue on Linux (I think because my Python is v2.4 and it seems hard to get it updated..) Cheers Andy On further investigation, it's actually not vary hard to update to Python 2.5 (well, run 2.5 alongside 2.4). What is hard is getting the Python ElementTree package to run happily within a Ruby environment. This is the case in the RestYahoo sample where we have a Ruby local client invoking a Python component. I get an exception thrown within the ElementTree code when it tries to load the xml.parsers.expat module (which is a binary library). I think this is due to Python needing to use its newer version of the expat lib than the one that Ruby must have loaded. There doesn't seem to be a way of upgrading the expat package on my system - the 1.95 version I have is required by 400 other packages! We could change the sample to have a Python client instead of Ruby, as it is more about demonstrating the REST extension rather than Ruby talking to Python, but it looks like we have a restriction here. Anyone have any ideas? Not off the top of my head. You could try building Ruby against the expat that Python needs??? Cheers, -- Pete
Re: Native M3 release status
I now have Ruby working on Mac. I will remove the ws bindings and clients from the *Calculator samples so we have a simple sample for each language. $TUSCANY_SCACPP/extensions/ruby/lib needs to be added to the LD_LIBRARY_PATH (or PATH on windows) to run the Ruby clients. I've updated the runclient scripts. The Rest binding builds on Mac but is crashing when the SCARuntime loads the extension which tries to register the reference binding extension. I'll investigate that next. Cheers, -- Pete
Re: Native M3 release status
On 2/21/07, Pete Robbins [EMAIL PROTECTED] wrote: I now have Ruby working on Mac. I will remove the ws bindings and clients from the *Calculator samples so we have a simple sample for each language. $TUSCANY_SCACPP/extensions/ruby/lib needs to be added to the LD_LIBRARY_PATH (or PATH on windows) to run the Ruby clients. I've updated the runclient scripts. Running a Linux build against the latest code gives me an error with the new Ruby layout: make[5]: Entering directory `/home/ajborley/workspace/TuscanyCPP/sca/runtime/extensions/ruby/extension/src' g++ -fPIC -O2 -g -pipe -Wall -Wp,-D_FORTIFY_SOURCE=2 -fexceptions -fstack-protector --param=ssp-buffer-size=4 -m32 -march=i386 -mtune=generic -fasynchronous-unwind-tables -Wall -fPIC -I. -I/usr/lib/ruby/1.8/i386-linux -I/usr/lib/ruby/1.8/i386-linux -I. -c Extension.cpp Extension.cpp:28:35: error: tuscany/sca/ruby/Ruby.h: No such file or directory Extension.cpp:29:51: error: tuscany/sca/ruby/RubyCompositeContext.h: No such file or directory Did everything get checked in? Am I missing an env variable? Cheers Andy - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
Re: Native M3 release status
On 2/21/07, Pete Robbins [EMAIL PROTECTED] wrote: On 21/02/07, Andrew Borley [EMAIL PROTECTED] wrote: On 2/21/07, Pete Robbins [EMAIL PROTECTED] wrote: I now have Ruby working on Mac. I will remove the ws bindings and clients from the *Calculator samples so we have a simple sample for each language. $TUSCANY_SCACPP/extensions/ruby/lib needs to be added to the LD_LIBRARY_PATH (or PATH on windows) to run the Ruby clients. I've updated the runclient scripts. Running a Linux build against the latest code gives me an error with the new Ruby layout: make[5]: Entering directory `/home/ajborley/workspace/TuscanyCPP/sca/runtime/extensions/ruby/extension/src' g++ -fPIC -O2 -g -pipe -Wall -Wp,-D_FORTIFY_SOURCE=2 -fexceptions -fstack-protector --param=ssp-buffer-size=4 -m32 -march=i386 -mtune=generic -fasynchronous-unwind-tables -Wall -fPIC -I. -I/usr/lib/ruby/1.8/i386-linux -I/usr/lib/ruby/1.8/i386-linux -I. -c Extension.cpp Extension.cpp:28:35: error: tuscany/sca/ruby/Ruby.h: No such file or directory Extension.cpp:29:51: error: tuscany/sca/ruby/RubyCompositeContext.h: No such file or directory Did everything get checked in? Am I missing an env variable? Everything is checked in. The extconf.rb RUby script generates the makefile. It should add the lib and include paths nexessary. Are there messages before the compile like: checking for #include tuscany/sca/ruby/RubyCompositeContext.h ? Yep, the full build log for sca/runtime/extensions/ruby/extension is: Making install in extension make[4]: Entering directory `/home/ajborley/workspace/TuscanyCPP/sca/runtime/extensions/ruby/extension' cd src; ruby extconf.rb; make checking for tuscany/sca/ruby/RubyCompositeContext.h... yes checking for main() in -ltuscany_sca_ruby_lang... yes creating Makefile make[5]: Entering directory `/home/ajborley/workspace/TuscanyCPP/sca/runtime/extensions/ruby/extension/src' g++ -fPIC -O2 -g -pipe -Wall -Wp,-D_FORTIFY_SOURCE=2 -fexceptions -fstack-protector --param=ssp-buffer-size=4 -m32 -march=i386 -mtune=generic -fasynchronous-unwind-tables -Wall -fPIC -I. -I/usr/lib/ruby/1.8/i386-linux -I/usr/lib/ruby/1.8/i386-linux -I. -c Extension.cpp Extension.cpp:28:35: error: tuscany/sca/ruby/Ruby.h: No such file or directory Extension.cpp:29:51: error: tuscany/sca/ruby/RubyCompositeContext.h: No such file or directory Extension.cpp:35: error: âVALUEâ does not name a type Extension.cpp: In function âvoid Init_tuscany_sca_ruby()â: Extension.cpp:47: error: âVALUEâ was not declared in this scope Extension.cpp:47: error: expected `;' before âmoduleâ Extension.cpp:48: error: âmoduleâ was not declared in this scope Extension.cpp:48: error: expected primary-expression before â)â token Extension.cpp:48: error: âANYARGSâ was not declared in this scope Extension.cpp:48: error: ârb_define_module_functionâ was not declared in this scope make[5]: *** [Extension.o] Error 1 make[5]: Leaving directory `/home/ajborley/workspace/TuscanyCPP/sca/runtime/extensions/ruby/extension/src' make[4]: *** [extension_build] Error 2 make[4]: Leaving directory `/home/ajborley/workspace/TuscanyCPP/sca/runtime/extensions/ruby/extension' make[3]: *** [install-recursive] Error 1 make[3]: Leaving directory `/home/ajborley/workspace/TuscanyCPP/sca/runtime/extensions/ruby' make[2]: *** [install-recursive] Error 1 make[2]: Leaving directory `/home/ajborley/workspace/TuscanyCPP/sca/runtime/extensions' make[1]: *** [install-recursive] Error 1 make[1]: Leaving directory `/home/ajborley/workspace/TuscanyCPP/sca/runtime' make: *** [install-recursive] Error 1 So the check for tuscany/sca/ruby/RubyCompositeContext.h is there (although there isn't one for Ruby.h) and looks like it succeeds, but the build still seems to fail to find it. Cheers Andy - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
Re: Native M3 release status
On 21/02/07, Andrew Borley [EMAIL PROTECTED] wrote: On 2/21/07, Pete Robbins [EMAIL PROTECTED] wrote: On 21/02/07, Andrew Borley [EMAIL PROTECTED] wrote: On 2/21/07, Pete Robbins [EMAIL PROTECTED] wrote: I now have Ruby working on Mac. I will remove the ws bindings and clients from the *Calculator samples so we have a simple sample for each language. $TUSCANY_SCACPP/extensions/ruby/lib needs to be added to the LD_LIBRARY_PATH (or PATH on windows) to run the Ruby clients. I've updated the runclient scripts. Running a Linux build against the latest code gives me an error with the new Ruby layout: make[5]: Entering directory `/home/ajborley/workspace/TuscanyCPP/sca/runtime/extensions/ruby/extension/src' g++ -fPIC -O2 -g -pipe -Wall -Wp,-D_FORTIFY_SOURCE=2 -fexceptions -fstack-protector --param=ssp-buffer-size=4 -m32 -march=i386 -mtune=generic -fasynchronous-unwind-tables -Wall -fPIC -I. -I/usr/lib/ruby/1.8/i386-linux -I/usr/lib/ruby/1.8/i386-linux -I. -c Extension.cpp Extension.cpp:28:35: error: tuscany/sca/ruby/Ruby.h: No such file or directory Extension.cpp:29:51: error: tuscany/sca/ruby/RubyCompositeContext.h: No such file or directory Did everything get checked in? Am I missing an env variable? Everything is checked in. The extconf.rb RUby script generates the makefile. It should add the lib and include paths nexessary. Are there messages before the compile like: checking for #include tuscany/sca/ruby/RubyCompositeContext.h ? Yep, the full build log for sca/runtime/extensions/ruby/extension is: Making install in extension make[4]: Entering directory `/home/ajborley/workspace/TuscanyCPP/sca/runtime/extensions/ruby/extension' cd src; ruby extconf.rb; make checking for tuscany/sca/ruby/RubyCompositeContext.h... yes checking for main() in -ltuscany_sca_ruby_lang... yes creating Makefile make[5]: Entering directory `/home/ajborley/workspace/TuscanyCPP/sca/runtime/extensions/ruby/extension/src' g++ -fPIC -O2 -g -pipe -Wall -Wp,-D_FORTIFY_SOURCE=2 -fexceptions -fstack-protector --param=ssp-buffer-size=4 -m32 -march=i386 -mtune=generic -fasynchronous-unwind-tables -Wall -fPIC -I. -I/usr/lib/ruby/1.8/i386-linux -I/usr/lib/ruby/1.8/i386-linux -I. -c Extension.cpp Extension.cpp:28:35: error: tuscany/sca/ruby/Ruby.h: No such file or directory Extension.cpp:29:51: error: tuscany/sca/ruby/RubyCompositeContext.h: No such file or directory Extension.cpp:35: error: âVALUEâ does not name a type Extension.cpp: In function âvoid Init_tuscany_sca_ruby()â: Extension.cpp:47: error: âVALUEâ was not declared in this scope Extension.cpp:47: error: expected `;' before âmoduleâ Extension.cpp:48: error: âmoduleâ was not declared in this scope Extension.cpp:48: error: expected primary-expression before â)â token Extension.cpp:48: error: âANYARGSâ was not declared in this scope Extension.cpp:48: error: ârb_define_module_functionâ was not declared in this scope make[5]: *** [Extension.o] Error 1 make[5]: Leaving directory `/home/ajborley/workspace/TuscanyCPP/sca/runtime/extensions/ruby/extension/src' make[4]: *** [extension_build] Error 2 make[4]: Leaving directory `/home/ajborley/workspace/TuscanyCPP/sca/runtime/extensions/ruby/extension' make[3]: *** [install-recursive] Error 1 make[3]: Leaving directory `/home/ajborley/workspace/TuscanyCPP/sca/runtime/extensions/ruby' make[2]: *** [install-recursive] Error 1 make[2]: Leaving directory `/home/ajborley/workspace/TuscanyCPP/sca/runtime/extensions' make[1]: *** [install-recursive] Error 1 make[1]: Leaving directory `/home/ajborley/workspace/TuscanyCPP/sca/runtime' make: *** [install-recursive] Error 1 So the check for tuscany/sca/ruby/RubyCompositeContext.h is there (although there isn't one for Ruby.h) and looks like it succeeds, but the build still seems to fail to find it. The checks are supposed to add the -I and -L stuff for where it finds the headers and libraries so here should be a -I../../src in the command line. This works on Mac and my RHEL3 linux. Which linux are you running? -- Pete - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
Native M3 release status
I'm currently working through building/testing the extensions and samples across Windows, Linux and Mac OS. Cpp: extension and samples look good (CppBigBank does not run on Mac as we have no ws binding extension) Ruby: Works fine on Windows and Linux (after fix I applied yesterday). Currently re-structuring the extension to work with Mac, I think this should be done by tomorrow. Python: Basic sample runs fine across all platforms. PHP: ? - I haven't built this yet REST: partially tested on Win and Linux So... there's still a fair bit to get through. Anyone care to fill in some of the blanks? Cheers, -- Pete
Re: Native M3 release status
On 2/20/07, Pete Robbins [EMAIL PROTECTED] wrote: I'm currently working through building/testing the extensions and samples across Windows, Linux and Mac OS. Cpp: extension and samples look good (CppBigBank does not run on Mac as we have no ws binding extension) Ruby: Works fine on Windows and Linux (after fix I applied yesterday). Currently re-structuring the extension to work with Mac, I think this should be done by tomorrow. Python: Basic sample runs fine across all platforms. PHP: ? - I haven't built this yet REST: partially tested on Win and Linux So... there's still a fair bit to get through. Anyone care to fill in some of the blanks? Cheers, -- Pete All the WS, REST and Python samples run fine on my Windows system, but I'm having a Python issue on Linux (I think because my Python is v2.4 and it seems hard to get it updated..) Cheers Andy - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]