Re: Binary plugs download
It a good point especially now that the popularity of OpenJDK is soaring and more and more ports are being done. On Wed, Nov 24, 2010 at 9:15 PM, kevin diggs wrote: > The section of the page: > > http://download.java.net/openjdk/jdk7/ > > for downloading has a heading that says: > > Linux platform > > If this is for x86 only then please label it as such (more than having > i586 buried in the file/link name). > > Linux actually runs on quite a lot of different hardware (I got me an > ultra2 that can run Linux; a bunch of PowerMacs too. Even a DEC Alpha. > That would be a real hoot getting a JDK on that thing. Oh I forgot > about the 040 based Quadra 700.). > > Sorry for the rant. But this kinda pushes my buttons. > > Thanks! > > kevin >
Re: PowerPC build ???
AFAIK there isa PowerPC version of RHEL and the IcedTea flavor of OpenJDK likely runs on that.
Re: Need reviewer: NONFCS_BUILD_INFO to add to the non-fcs version string
Hi Kelly, Logic looks fine, just a nit Unless $(NONFCS_BUILD_INFO) really needs that leading "-", I'd remove it and add a "-" in line 110 and 297 below. Makes it a bit more readable Dave hotspot/make/defs.make 109 else 110 HOTSPOT_BUILD_VERSION=internal$(NONFCS_BUILD_INFO) 111 endif jdk/make/common/shared/Defs.gmk 290 ifndef NONFCS_BUILD_INFO 291 BUILD_DATE := $(shell $(DATE) '+%Y_%m_%d_%H_%M') 292 CLEAN_USERNAME := $(shell $(ECHO) "$(USER)" | $(TR) -d -c '[:alnum:]') 293 USER_RELEASE_SUFFIX := $(shell $(ECHO) "$(CLEAN_USERNAME)_$(BUILD_DATE)" | $(TR) '[:upper:]' '[:lower:]' ) 294 NONFCS_BUILD_INFO = -$(USER_RELEASE_SUFFIX) 295 endif 296 export NONFCS_BUILD_INFO 297 FULL_VERSION = $(RELEASE)$(NONFCS_BUILD_INFO)-$(BUILD_NUMBER) 298 endif On 11/24/2010 11:36 AM, Kelly O'Hair wrote: Dang... just shoot me now. :^( Try this: http://cr.openjdk.java.net/~ohair/openjdk7/nonfcs-version/webrev/ Sorry about that. -kto On Nov 24, 2010, at 11:28 AM, Mark Wielaard wrote: On Wed, 2010-11-24 at 11:19 -0800, Kelly O'Hair wrote: I need a reviewer for this change: 6987107: Add NONFCS_BUILD_INFO variable to add to but not modify MILESTONE in version string http://javaweb.sfbay.sun.com/~ohair/webrevs/jdk7/nonfcs-version/webrev/ Bit hard to review if the host isn't reachable :) Could you just attach the patch to your email, or post it on some publicly reachable machine? Thanks, Mark
Re: PowerPC build ???
On Wed, Nov 24, 2010 at 8:21 PM, Kelly O'Hair wrote: > Kevin, > > These blogs might help explain things: > http://blogs.sun.com/kto/entry/anatomy_of_the_jdk_build > http://blogs.sun.com/jjg/entry/building_javac_for_jdk7 > Thanks for the links. I'll look at these later. If they answer the following questions, sorry. > In jdk7/hotspot the BOOT jdk is used to do javac compilers, for some > serviceability tools, "serviceability tools": What exactly is this? The aforementioned jar and javah. > but that should probably be changed to use the BOOT jdk java with the > javac.jar from the > langtools repository. > "langtools repository"??? Is this in the source or "boot" jdk??? > -kto > > First, thanks for taking the time to reply. This question is a little ... weird ... but ... Is the thought of trying to build your sdk with gcj ... heresy (sp?). If not please give me ... your thought as to whether it has any chance of working. I would think that if it is for compiles and the like ... it might work ... Also, what is in this binary plug? Am I completely wasting my time since I don't have this? Any chance of getting a powerpc binary plug (or access to the source so one can be built)? This question will show my ignorance: What is an i586 jar file? I thought a jar file contained class files that would be platform independent? Is there native code in there to (too? - sorry ain't not no english major)? kevin
Re: PowerPC build ???
Kevin, These blogs might help explain things: http://blogs.sun.com/kto/entry/anatomy_of_the_jdk_build http://blogs.sun.com/jjg/entry/building_javac_for_jdk7 In jdk7/hotspot the BOOT jdk is used to do javac compilers, for some serviceability tools, but that should probably be changed to use the BOOT jdk java with the javac.jar from the langtools repository. -kto
Fwd: dead link
Hi, The FAQ link at the bottom of page: http://download.java.net/openjdk/jdk7/ is dead??? kevin
Re: Need reviewer: NONFCS_BUILD_INFO to add to the non-fcs version string
On Nov 24, 2010, at 6:06 PM, David Holmes wrote: Kelly O'Hair said the following on 11/25/10 11:38: On Nov 24, 2010, at 4:08 PM, David Holmes wrote: Ok. So why not just USER_BUILD_INFO or CUSTOM_BUILD_INFO ? There's really no tie to "FCS" or not. In fact why not use USER_RELEASE_SUFFIX - see below .. It's just a name, but we were looking for a name that told you it would be ignored if MILESTONE=fcs. I see - that's not obvious from the webrev. I guess somewhere in there is some "magic" MILESTONE treatment. That said for a real fcs build I expect the complete version strings to specified by the "release" process on the build invocation not defaulting to whatever is in the Makefiles. The whole way the versions are defined is a bit confusing, tricky stuff. Probably need a serious revamping someday. But in general, yes, if MILESTONE=fcs most values come from the command line or RE scripts. USER_RELEASE_SUFFIX would work, and your suggestion below works too, although I still want to export USER_RELEASE_SUFFIX so that shell logic isn't repeated over and over, hard on windows systems. I'm still unclear to where you are exporting this value? One of the doc makefiles utilizes USER_RELEASE_SUFFIX but doesn't need it to be exported presently. The jdk makefiles are heavily nested, make running make, and these Defs files are repeatedly parsed over and over, maybe 100's of times for a top level Makefile? The export pushes the variable into the environment so that the lower nested makefiles see the definition as an environment variable and if you use ifndef then you avoid the recalculations or shell execs. Solaris/Linux aren't bothered by the excessive execs that much, but Windows suffers from it. I've been trying to lower the total exec count when the Defs files are parsed to make windows builds faster. I'll need to change all NONFCS_BUILD_INFO to USER_RELEASE_SUFFIX in the hotspot files. That also cures John's complaint about the "fcs" name too. Check this one out: http://cr.openjdk.java.net/~ohair/openjdk7/nonfcs-version2/webrev/ I can give Thumbs Up as is. OK. -kto Thanks, David -kto When full builds of jdk7 are done, we wanted a single variable that would add some extra identification string to all version strings. Hudson systems and JPRT can use this to uniquely identify all full builds so that testing teams can report an exact build on bugs. I see. I'm not sure why you re-export the variable here (jdk/.../ Defs.gmk) as it doesn't appear to get used anywhere else: + ifndef NONFCS_BUILD_INFO BUILD_DATE := $(shell $(DATE) '+%Y_%m_%d_%H_%M') CLEAN_USERNAME := $(shell $(ECHO) "$(USER)" | $(TR) -d -c '[:alnum:]') USER_RELEASE_SUFFIX := $(shell $(ECHO) "$(CLEAN_USERNAME)_$ (BUILD_DATE)" | $(TR) '[:upper:]' '[:lower:]' ) ! NONFCS_BUILD_INFO = $(USER_RELEASE_SUFFIX) ! endif ! export NONFCS_BUILD_INFO ! FULL_VERSION = $(RELEASE)-$(NONFCS_BUILD_INFO)-$(BUILD_NUMBER) It would seem a little simpler/cleaner to me to instead just do: ifndef NONFCS_BUILD_INFO BUILD_DATE := $(shell $(DATE) '+%Y_%m_%d_%H_%M') CLEAN_USERNAME := $(shell $(ECHO) "$(USER)" | $(TR) -d -c '[:alnum:]') USER_RELEASE_SUFFIX := $(shell $(ECHO) "$(CLEAN_USERNAME)_$ (BUILD_DATE)" | $(TR) '[:upper:]' '[:lower:]' ) else USER_RELEASE_SUFFIX := $(NONFCS_BUILD_INFO) endif FULL_VERSION = $(RELEASE)-$(USER_RELEASE_SUFFIX)-$(BUILD_NUMBER) And this would be even simpler if USER_RELEASE_SUFFIX were the variable that was set externally in the first place. ifndef USER_RELEASE_SUFFIX BUILD_DATE := $(shell $(DATE) '+%Y_%m_%d_%H_%M') CLEAN_USERNAME := $(shell $(ECHO) "$(USER)" | $(TR) -d -c '[:alnum:]') USER_RELEASE_SUFFIX := $(shell $(ECHO) "$(CLEAN_USERNAME)_$ (BUILD_DATE)" | $(TR) '[:upper:]' '[:lower:]' ) endif FULL_VERSION = $(RELEASE)-$(USER_RELEASE_SUFFIX)-$(BUILD_NUMBER) Cheers, David Any why use the same NONFCS_BUILD_INFO for Hotspot and the JDK when they typically report different version strings anyway ??? It's extra build information, not really a change in the component version number. -kto David Aside from that, looks fine. -John On Nov 24, 2010, at 11:28 AM, Mark Wielaard wrote: On Wed, 2010-11-24 at 11:19 -0800, Kelly O'Hair wrote: I need a reviewer for this change: 6987107: Add NONFCS_BUILD_INFO variable to add to but not modify MILESTONE in version string http://javaweb.sfbay.sun.com/~ohair/webrevs/jdk7/nonfcs-version/webrev/ Bit hard to review if the host isn't reachable :) Could you just attach the patch to your email, or post it on some publicly reachable machine? Thanks, Mark
Fwd: Binary plugs download
The section of the page: http://download.java.net/openjdk/jdk7/ for downloading has a heading that says: Linux platform If this is for x86 only then please label it as such (more than having i586 buried in the file/link name). Linux actually runs on quite a lot of different hardware (I got me an ultra2 that can run Linux; a bunch of PowerMacs too. Even a DEC Alpha. That would be a real hoot getting a JDK on that thing. Oh I forgot about the 040 based Quadra 700.). Sorry for the rant. But this kinda pushes my buttons. Thanks! kevin
Re: PowerPC build ???
kevin diggs said the following on 11/25/10 11:43: Is there any way to build this thing on a Linux PowerPC box (YDL 4.0 & 6.0). Is there a snowballs's chance in some place hot that GCC gcj will work? I admit that I have not done an extensive search for a PowerPC Linux jdk (May actually be one on one of the install CDs but I doubt it). Maybe on IBM's site somewhere? I do NOT define Open and run everywhere as Windows and/or x86. Sorry don't know what the status of an open PPC implementation is but ... P.S.: What is the "bootstrap jdk" used for? Access to a java compiler? Or does it use other "stuff" from in there? It will supply javac, javah, and jar at least. David Holmes
Re: Need reviewer: NONFCS_BUILD_INFO to add to the non-fcs version string
Kelly O'Hair said the following on 11/25/10 11:38: On Nov 24, 2010, at 4:08 PM, David Holmes wrote: Ok. So why not just USER_BUILD_INFO or CUSTOM_BUILD_INFO ? There's really no tie to "FCS" or not. In fact why not use USER_RELEASE_SUFFIX - see below .. It's just a name, but we were looking for a name that told you it would be ignored if MILESTONE=fcs. I see - that's not obvious from the webrev. I guess somewhere in there is some "magic" MILESTONE treatment. That said for a real fcs build I expect the complete version strings to specified by the "release" process on the build invocation not defaulting to whatever is in the Makefiles. USER_RELEASE_SUFFIX would work, and your suggestion below works too, although I still want to export USER_RELEASE_SUFFIX so that shell logic isn't repeated over and over, hard on windows systems. I'm still unclear to where you are exporting this value? One of the doc makefiles utilizes USER_RELEASE_SUFFIX but doesn't need it to be exported presently. I'll need to change all NONFCS_BUILD_INFO to USER_RELEASE_SUFFIX in the hotspot files. That also cures John's complaint about the "fcs" name too. Check this one out: http://cr.openjdk.java.net/~ohair/openjdk7/nonfcs-version2/webrev/ I can give Thumbs Up as is. Thanks, David -kto When full builds of jdk7 are done, we wanted a single variable that would add some extra identification string to all version strings. Hudson systems and JPRT can use this to uniquely identify all full builds so that testing teams can report an exact build on bugs. I see. I'm not sure why you re-export the variable here (jdk/.../Defs.gmk) as it doesn't appear to get used anywhere else: + ifndef NONFCS_BUILD_INFO BUILD_DATE := $(shell $(DATE) '+%Y_%m_%d_%H_%M') CLEAN_USERNAME := $(shell $(ECHO) "$(USER)" | $(TR) -d -c '[:alnum:]') USER_RELEASE_SUFFIX := $(shell $(ECHO) "$(CLEAN_USERNAME)_$(BUILD_DATE)" | $(TR) '[:upper:]' '[:lower:]' ) ! NONFCS_BUILD_INFO = $(USER_RELEASE_SUFFIX) ! endif ! export NONFCS_BUILD_INFO ! FULL_VERSION = $(RELEASE)-$(NONFCS_BUILD_INFO)-$(BUILD_NUMBER) It would seem a little simpler/cleaner to me to instead just do: ifndef NONFCS_BUILD_INFO BUILD_DATE := $(shell $(DATE) '+%Y_%m_%d_%H_%M') CLEAN_USERNAME := $(shell $(ECHO) "$(USER)" | $(TR) -d -c '[:alnum:]') USER_RELEASE_SUFFIX := $(shell $(ECHO) "$(CLEAN_USERNAME)_$(BUILD_DATE)" | $(TR) '[:upper:]' '[:lower:]' ) else USER_RELEASE_SUFFIX := $(NONFCS_BUILD_INFO) endif FULL_VERSION = $(RELEASE)-$(USER_RELEASE_SUFFIX)-$(BUILD_NUMBER) And this would be even simpler if USER_RELEASE_SUFFIX were the variable that was set externally in the first place. ifndef USER_RELEASE_SUFFIX BUILD_DATE := $(shell $(DATE) '+%Y_%m_%d_%H_%M') CLEAN_USERNAME := $(shell $(ECHO) "$(USER)" | $(TR) -d -c '[:alnum:]') USER_RELEASE_SUFFIX := $(shell $(ECHO) "$(CLEAN_USERNAME)_$(BUILD_DATE)" | $(TR) '[:upper:]' '[:lower:]' ) endif FULL_VERSION = $(RELEASE)-$(USER_RELEASE_SUFFIX)-$(BUILD_NUMBER) Cheers, David Any why use the same NONFCS_BUILD_INFO for Hotspot and the JDK when they typically report different version strings anyway ??? It's extra build information, not really a change in the component version number. -kto David Aside from that, looks fine. -John On Nov 24, 2010, at 11:28 AM, Mark Wielaard wrote: On Wed, 2010-11-24 at 11:19 -0800, Kelly O'Hair wrote: I need a reviewer for this change: 6987107: Add NONFCS_BUILD_INFO variable to add to but not modify MILESTONE in version string http://javaweb.sfbay.sun.com/~ohair/webrevs/jdk7/nonfcs-version/webrev/ Bit hard to review if the host isn't reachable :) Could you just attach the patch to your email, or post it on some publicly reachable machine? Thanks, Mark
PowerPC build ???
Hi, Is there any way to build this thing on a Linux PowerPC box (YDL 4.0 & 6.0). Is there a snowballs's chance in some place hot that GCC gcj will work? I admit that I have not done an extensive search for a PowerPC Linux jdk (May actually be one on one of the install CDs but I doubt it). Maybe on IBM's site somewhere? I do NOT define Open and run everywhere as Windows and/or x86. Thanks! kevin P.S.: What is the "bootstrap jdk" used for? Access to a java compiler? Or does it use other "stuff" from in there?
Re: Need reviewer: NONFCS_BUILD_INFO to add to the non-fcs version string
On Nov 24, 2010, at 4:08 PM, David Holmes wrote: Kelly O'Hair said the following on 11/25/10 09:55: On Nov 24, 2010, at 3:47 PM, David Holmes wrote: John Coomes said the following on 11/25/10 09:16: Kelly O'Hair (kelly.oh...@oracle.com) wrote: Dang... just shoot me now. :^( Try this: http://cr.openjdk.java.net/~ohair/openjdk7/nonfcs-version/webrev/ Sorry about that. Aren't we using the term GA (general availability) instead of FCS these days? Going further why is this even an issue? In all interesting cases HOTSPOT_BUILD_VERSION should be set on the command-line. The Makefile only needs to give a default if it is not set. So why set NONFCS_BUILD_INFO when you can set HOTSPOT_BUILD_VERSION in the first place? I'm trying to avoid having to specify a specific variable for every component of the jdk when we want to add specific build information to the version string. Ok. So why not just USER_BUILD_INFO or CUSTOM_BUILD_INFO ? There's really no tie to "FCS" or not. In fact why not use USER_RELEASE_SUFFIX - see below .. It's just a name, but we were looking for a name that told you it would be ignored if MILESTONE=fcs. USER_RELEASE_SUFFIX would work, and your suggestion below works too, although I still want to export USER_RELEASE_SUFFIX so that shell logic isn't repeated over and over, hard on windows systems. I'll need to change all NONFCS_BUILD_INFO to USER_RELEASE_SUFFIX in the hotspot files. That also cures John's complaint about the "fcs" name too. Check this one out: http://cr.openjdk.java.net/~ohair/openjdk7/nonfcs-version2/webrev/ -kto When full builds of jdk7 are done, we wanted a single variable that would add some extra identification string to all version strings. Hudson systems and JPRT can use this to uniquely identify all full builds so that testing teams can report an exact build on bugs. I see. I'm not sure why you re-export the variable here (jdk/.../ Defs.gmk) as it doesn't appear to get used anywhere else: + ifndef NONFCS_BUILD_INFO BUILD_DATE := $(shell $(DATE) '+%Y_%m_%d_%H_%M') CLEAN_USERNAME := $(shell $(ECHO) "$(USER)" | $(TR) -d -c '[:alnum:]') USER_RELEASE_SUFFIX := $(shell $(ECHO) "$(CLEAN_USERNAME)_$ (BUILD_DATE)" | $(TR) '[:upper:]' '[:lower:]' ) ! NONFCS_BUILD_INFO = $(USER_RELEASE_SUFFIX) ! endif ! export NONFCS_BUILD_INFO ! FULL_VERSION = $(RELEASE)-$(NONFCS_BUILD_INFO)-$(BUILD_NUMBER) It would seem a little simpler/cleaner to me to instead just do: ifndef NONFCS_BUILD_INFO BUILD_DATE := $(shell $(DATE) '+%Y_%m_%d_%H_%M') CLEAN_USERNAME := $(shell $(ECHO) "$(USER)" | $(TR) -d -c '[:alnum:]') USER_RELEASE_SUFFIX := $(shell $(ECHO) "$(CLEAN_USERNAME)_$ (BUILD_DATE)" | $(TR) '[:upper:]' '[:lower:]' ) else USER_RELEASE_SUFFIX := $(NONFCS_BUILD_INFO) endif FULL_VERSION = $(RELEASE)-$(USER_RELEASE_SUFFIX)-$(BUILD_NUMBER) And this would be even simpler if USER_RELEASE_SUFFIX were the variable that was set externally in the first place. ifndef USER_RELEASE_SUFFIX BUILD_DATE := $(shell $(DATE) '+%Y_%m_%d_%H_%M') CLEAN_USERNAME := $(shell $(ECHO) "$(USER)" | $(TR) -d -c '[:alnum:]') USER_RELEASE_SUFFIX := $(shell $(ECHO) "$(CLEAN_USERNAME)_$ (BUILD_DATE)" | $(TR) '[:upper:]' '[:lower:]' ) endif FULL_VERSION = $(RELEASE)-$(USER_RELEASE_SUFFIX)-$(BUILD_NUMBER) Cheers, David Any why use the same NONFCS_BUILD_INFO for Hotspot and the JDK when they typically report different version strings anyway ??? It's extra build information, not really a change in the component version number. -kto David Aside from that, looks fine. -John On Nov 24, 2010, at 11:28 AM, Mark Wielaard wrote: On Wed, 2010-11-24 at 11:19 -0800, Kelly O'Hair wrote: I need a reviewer for this change: 6987107: Add NONFCS_BUILD_INFO variable to add to but not modify MILESTONE in version string http://javaweb.sfbay.sun.com/~ohair/webrevs/jdk7/nonfcs-version/webrev/ Bit hard to review if the host isn't reachable :) Could you just attach the patch to your email, or post it on some publicly reachable machine? Thanks, Mark
Re: Need reviewer: NONFCS_BUILD_INFO to add to the non-fcs version string
Kelly O'Hair said the following on 11/25/10 09:55: On Nov 24, 2010, at 3:47 PM, David Holmes wrote: John Coomes said the following on 11/25/10 09:16: Kelly O'Hair (kelly.oh...@oracle.com) wrote: Dang... just shoot me now. :^( Try this: http://cr.openjdk.java.net/~ohair/openjdk7/nonfcs-version/webrev/ Sorry about that. Aren't we using the term GA (general availability) instead of FCS these days? Going further why is this even an issue? In all interesting cases HOTSPOT_BUILD_VERSION should be set on the command-line. The Makefile only needs to give a default if it is not set. So why set NONFCS_BUILD_INFO when you can set HOTSPOT_BUILD_VERSION in the first place? I'm trying to avoid having to specify a specific variable for every component of the jdk when we want to add specific build information to the version string. Ok. So why not just USER_BUILD_INFO or CUSTOM_BUILD_INFO ? There's really no tie to "FCS" or not. In fact why not use USER_RELEASE_SUFFIX - see below .. When full builds of jdk7 are done, we wanted a single variable that would add some extra identification string to all version strings. Hudson systems and JPRT can use this to uniquely identify all full builds so that testing teams can report an exact build on bugs. I see. I'm not sure why you re-export the variable here (jdk/.../Defs.gmk) as it doesn't appear to get used anywhere else: + ifndef NONFCS_BUILD_INFO BUILD_DATE := $(shell $(DATE) '+%Y_%m_%d_%H_%M') CLEAN_USERNAME := $(shell $(ECHO) "$(USER)" | $(TR) -d -c '[:alnum:]') USER_RELEASE_SUFFIX := $(shell $(ECHO) "$(CLEAN_USERNAME)_$(BUILD_DATE)" | $(TR) '[:upper:]' '[:lower:]' ) ! NONFCS_BUILD_INFO = $(USER_RELEASE_SUFFIX) ! endif ! export NONFCS_BUILD_INFO ! FULL_VERSION = $(RELEASE)-$(NONFCS_BUILD_INFO)-$(BUILD_NUMBER) It would seem a little simpler/cleaner to me to instead just do: ifndef NONFCS_BUILD_INFO BUILD_DATE := $(shell $(DATE) '+%Y_%m_%d_%H_%M') CLEAN_USERNAME := $(shell $(ECHO) "$(USER)" | $(TR) -d -c '[:alnum:]') USER_RELEASE_SUFFIX := $(shell $(ECHO) "$(CLEAN_USERNAME)_$(BUILD_DATE)" | $(TR) '[:upper:]' '[:lower:]' ) else USER_RELEASE_SUFFIX := $(NONFCS_BUILD_INFO) endif FULL_VERSION = $(RELEASE)-$(USER_RELEASE_SUFFIX)-$(BUILD_NUMBER) And this would be even simpler if USER_RELEASE_SUFFIX were the variable that was set externally in the first place. ifndef USER_RELEASE_SUFFIX BUILD_DATE := $(shell $(DATE) '+%Y_%m_%d_%H_%M') CLEAN_USERNAME := $(shell $(ECHO) "$(USER)" | $(TR) -d -c '[:alnum:]') USER_RELEASE_SUFFIX := $(shell $(ECHO) "$(CLEAN_USERNAME)_$(BUILD_DATE)" | $(TR) '[:upper:]' '[:lower:]' ) endif FULL_VERSION = $(RELEASE)-$(USER_RELEASE_SUFFIX)-$(BUILD_NUMBER) Cheers, David Any why use the same NONFCS_BUILD_INFO for Hotspot and the JDK when they typically report different version strings anyway ??? It's extra build information, not really a change in the component version number. -kto David Aside from that, looks fine. -John On Nov 24, 2010, at 11:28 AM, Mark Wielaard wrote: On Wed, 2010-11-24 at 11:19 -0800, Kelly O'Hair wrote: I need a reviewer for this change: 6987107: Add NONFCS_BUILD_INFO variable to add to but not modify MILESTONE in version string http://javaweb.sfbay.sun.com/~ohair/webrevs/jdk7/nonfcs-version/webrev/ Bit hard to review if the host isn't reachable :) Could you just attach the patch to your email, or post it on some publicly reachable machine? Thanks, Mark
Re: Need reviewer: NONFCS_BUILD_INFO to add to the non-fcs version string
On Nov 24, 2010, at 3:47 PM, David Holmes wrote: John Coomes said the following on 11/25/10 09:16: Kelly O'Hair (kelly.oh...@oracle.com) wrote: Dang... just shoot me now. :^( Try this: http://cr.openjdk.java.net/~ohair/openjdk7/nonfcs-version/webrev/ Sorry about that. Aren't we using the term GA (general availability) instead of FCS these days? Going further why is this even an issue? In all interesting cases HOTSPOT_BUILD_VERSION should be set on the command-line. The Makefile only needs to give a default if it is not set. So why set NONFCS_BUILD_INFO when you can set HOTSPOT_BUILD_VERSION in the first place? I'm trying to avoid having to specify a specific variable for every component of the jdk when we want to add specific build information to the version string. When full builds of jdk7 are done, we wanted a single variable that would add some extra identification string to all version strings. Hudson systems and JPRT can use this to uniquely identify all full builds so that testing teams can report an exact build on bugs. Any why use the same NONFCS_BUILD_INFO for Hotspot and the JDK when they typically report different version strings anyway ??? It's extra build information, not really a change in the component version number. -kto David Aside from that, looks fine. -John On Nov 24, 2010, at 11:28 AM, Mark Wielaard wrote: On Wed, 2010-11-24 at 11:19 -0800, Kelly O'Hair wrote: I need a reviewer for this change: 6987107: Add NONFCS_BUILD_INFO variable to add to but not modify MILESTONE in version string http://javaweb.sfbay.sun.com/~ohair/webrevs/jdk7/nonfcs-version/webrev/ Bit hard to review if the host isn't reachable :) Could you just attach the patch to your email, or post it on some publicly reachable machine? Thanks, Mark
Re: Need reviewer: NONFCS_BUILD_INFO to add to the non-fcs version string
John Coomes said the following on 11/25/10 09:16: Kelly O'Hair (kelly.oh...@oracle.com) wrote: Dang... just shoot me now. :^( Try this: http://cr.openjdk.java.net/~ohair/openjdk7/nonfcs-version/webrev/ Sorry about that. Aren't we using the term GA (general availability) instead of FCS these days? Going further why is this even an issue? In all interesting cases HOTSPOT_BUILD_VERSION should be set on the command-line. The Makefile only needs to give a default if it is not set. So why set NONFCS_BUILD_INFO when you can set HOTSPOT_BUILD_VERSION in the first place? Any why use the same NONFCS_BUILD_INFO for Hotspot and the JDK when they typically report different version strings anyway ??? David Aside from that, looks fine. -John On Nov 24, 2010, at 11:28 AM, Mark Wielaard wrote: On Wed, 2010-11-24 at 11:19 -0800, Kelly O'Hair wrote: I need a reviewer for this change: 6987107: Add NONFCS_BUILD_INFO variable to add to but not modify MILESTONE in version string http://javaweb.sfbay.sun.com/~ohair/webrevs/jdk7/nonfcs-version/webrev/ Bit hard to review if the host isn't reachable :) Could you just attach the patch to your email, or post it on some publicly reachable machine? Thanks, Mark
Re: Need reviewer: NONFCS_BUILD_INFO to add to the non-fcs version string
On Nov 24, 2010, at 3:16 PM, John Coomes wrote: Kelly O'Hair (kelly.oh...@oracle.com) wrote: Dang... just shoot me now. :^( Try this: http://cr.openjdk.java.net/~ohair/openjdk7/nonfcs-version/webrev/ Sorry about that. Aren't we using the term GA (general availability) instead of FCS these days? I thought GA was Lady Gaga's first name? ;^) I haven't heard about any effort to purge FCS from the makefiles. Lots of changes would be needed for that. The term FCS seems to be generally recognized as the same thing as GA: http://en.wikipedia.org/wiki/Software_release_life_cycle#General_availability -kto Aside from that, looks fine. -John On Nov 24, 2010, at 11:28 AM, Mark Wielaard wrote: On Wed, 2010-11-24 at 11:19 -0800, Kelly O'Hair wrote: I need a reviewer for this change: 6987107: Add NONFCS_BUILD_INFO variable to add to but not modify MILESTONE in version string http://javaweb.sfbay.sun.com/~ohair/webrevs/jdk7/nonfcs-version/webrev/ Bit hard to review if the host isn't reachable :) Could you just attach the patch to your email, or post it on some publicly reachable machine? Thanks, Mark
Re: Need reviewer: NONFCS_BUILD_INFO to add to the non-fcs version string
Kelly O'Hair (kelly.oh...@oracle.com) wrote: > Dang... just shoot me now. :^( > > Try this: >http://cr.openjdk.java.net/~ohair/openjdk7/nonfcs-version/webrev/ > > Sorry about that. Aren't we using the term GA (general availability) instead of FCS these days? Aside from that, looks fine. -John > On Nov 24, 2010, at 11:28 AM, Mark Wielaard wrote: > > > On Wed, 2010-11-24 at 11:19 -0800, Kelly O'Hair wrote: > >> I need a reviewer for this change: > >> > >> 6987107: Add NONFCS_BUILD_INFO variable to add to but not modify > >> MILESTONE in version string > >> http://javaweb.sfbay.sun.com/~ohair/webrevs/jdk7/nonfcs-version/webrev/ > > > > Bit hard to review if the host isn't reachable :) > > Could you just attach the patch to your email, > > or post it on some publicly reachable machine? > > > > Thanks, > > > > Mark > > >
Re: Need reviewer: NONFCS_BUILD_INFO to add to the non-fcs version string
Suggestion accepted. Thanks. Check http://cr.openjdk.java.net/~ohair/openjdk7/nonfcs-version/webrev/ I also found a few additional places to correct information on HOTSPOT_BUILD_VERSION. If this looks ok I'll build on all systems and see how it looks. -kto On Nov 24, 2010, at 11:52 AM, David Katleman wrote: Hi Kelly, Logic looks fine, just a nit Unless $(NONFCS_BUILD_INFO) really needs that leading "-", I'd remove it and add a "-" in line 110 and 297 below. Makes it a bit more readable Dave hotspot/make/defs.make 109 else 110 HOTSPOT_BUILD_VERSION=internal$(NONFCS_BUILD_INFO) 111 endif jdk/make/common/shared/Defs.gmk 290 ifndef NONFCS_BUILD_INFO 291 BUILD_DATE := $(shell $(DATE) '+%Y_%m_%d_%H_%M') 292 CLEAN_USERNAME := $(shell $(ECHO) "$(USER)" | $(TR) -d -c '[:alnum:]') 293 USER_RELEASE_SUFFIX := $(shell $(ECHO) "$(CLEAN_USERNAME)_$ (BUILD_DATE)" | $(TR) '[:upper:]' '[:lower:]' ) 294 NONFCS_BUILD_INFO = -$(USER_RELEASE_SUFFIX) 295 endif 296 export NONFCS_BUILD_INFO 297 FULL_VERSION = $(RELEASE)$(NONFCS_BUILD_INFO)-$(BUILD_NUMBER) 298 endif On 11/24/2010 11:36 AM, Kelly O'Hair wrote: Dang... just shoot me now. :^( Try this: http://cr.openjdk.java.net/~ohair/openjdk7/nonfcs-version/webrev/ Sorry about that. -kto On Nov 24, 2010, at 11:28 AM, Mark Wielaard wrote: On Wed, 2010-11-24 at 11:19 -0800, Kelly O'Hair wrote: I need a reviewer for this change: 6987107: Add NONFCS_BUILD_INFO variable to add to but not modify MILESTONE in version string http://javaweb.sfbay.sun.com/~ohair/webrevs/jdk7/nonfcs-version/webrev/ Bit hard to review if the host isn't reachable :) Could you just attach the patch to your email, or post it on some publicly reachable machine? Thanks, Mark
Re: Need reviewer: NONFCS_BUILD_INFO to add to the non-fcs version string
On Nov 24, 2010, at 12:03 PM, Dr Andrew John Hughes wrote: On 11:36 Wed 24 Nov , Kelly O'Hair wrote: Dang... just shoot me now. :^( Try this: http://cr.openjdk.java.net/~ohair/openjdk7/nonfcs-version/webrev/ Sorry about that. Looks ok to me. I assume the motivation is so that MILESTONE doesn't have to be duplicated in setting JPRT_BUILD_VERSION? You can instead now just set NONFCS_BUILD_INFO to replace the autogenerated user/date stuff without getting rid of MILESTONE. Yup. Basically. We have various build systems doing non-fcs builds and I wanted a generic way to add to the version string on those builds so they can be identified. -kto -kto On Nov 24, 2010, at 11:28 AM, Mark Wielaard wrote: On Wed, 2010-11-24 at 11:19 -0800, Kelly O'Hair wrote: I need a reviewer for this change: 6987107: Add NONFCS_BUILD_INFO variable to add to but not modify MILESTONE in version string http://javaweb.sfbay.sun.com/~ohair/webrevs/jdk7/nonfcs-version/webrev/ Bit hard to review if the host isn't reachable :) Could you just attach the patch to your email, or post it on some publicly reachable machine? Thanks, Mark -- Andrew :) Free Java Software Engineer Red Hat, Inc. (http://www.redhat.com) Support Free Java! Contribute to GNU Classpath and the OpenJDK http://www.gnu.org/software/classpath http://openjdk.java.net PGP Key: 94EFD9D8 (http://subkeys.pgp.net) Fingerprint = F8EF F1EA 401E 2E60 15FA 7927 142C 2591 94EF D9D8
Building OpenJDK 7 under Fedora 14'
I finally got the build done with Fedora 14! Thanks all your hints.
Re: Need reviewer: NONFCS_BUILD_INFO to add to the non-fcs version string
On 11:36 Wed 24 Nov , Kelly O'Hair wrote: > Dang... just shoot me now. :^( > > Try this: >http://cr.openjdk.java.net/~ohair/openjdk7/nonfcs-version/webrev/ > > Sorry about that. > Looks ok to me. I assume the motivation is so that MILESTONE doesn't have to be duplicated in setting JPRT_BUILD_VERSION? You can instead now just set NONFCS_BUILD_INFO to replace the autogenerated user/date stuff without getting rid of MILESTONE. > -kto > > On Nov 24, 2010, at 11:28 AM, Mark Wielaard wrote: > > > On Wed, 2010-11-24 at 11:19 -0800, Kelly O'Hair wrote: > >> I need a reviewer for this change: > >> > >> 6987107: Add NONFCS_BUILD_INFO variable to add to but not modify > >> MILESTONE in version string > >> http://javaweb.sfbay.sun.com/~ohair/webrevs/jdk7/nonfcs-version/webrev/ > > > > Bit hard to review if the host isn't reachable :) > > Could you just attach the patch to your email, > > or post it on some publicly reachable machine? > > > > Thanks, > > > > Mark > > > -- Andrew :) Free Java Software Engineer Red Hat, Inc. (http://www.redhat.com) Support Free Java! Contribute to GNU Classpath and the OpenJDK http://www.gnu.org/software/classpath http://openjdk.java.net PGP Key: 94EFD9D8 (http://subkeys.pgp.net) Fingerprint = F8EF F1EA 401E 2E60 15FA 7927 142C 2591 94EF D9D8
Re: Need reviewer: NONFCS_BUILD_INFO to add to the non-fcs version string
Dang... just shoot me now. :^( Try this: http://cr.openjdk.java.net/~ohair/openjdk7/nonfcs-version/webrev/ Sorry about that. -kto On Nov 24, 2010, at 11:28 AM, Mark Wielaard wrote: On Wed, 2010-11-24 at 11:19 -0800, Kelly O'Hair wrote: I need a reviewer for this change: 6987107: Add NONFCS_BUILD_INFO variable to add to but not modify MILESTONE in version string http://javaweb.sfbay.sun.com/~ohair/webrevs/jdk7/nonfcs-version/webrev/ Bit hard to review if the host isn't reachable :) Could you just attach the patch to your email, or post it on some publicly reachable machine? Thanks, Mark
Re: Need reviewer: NONFCS_BUILD_INFO to add to the non-fcs version string
On Wed, 2010-11-24 at 11:19 -0800, Kelly O'Hair wrote: > I need a reviewer for this change: > > 6987107: Add NONFCS_BUILD_INFO variable to add to but not modify > MILESTONE in version string > http://javaweb.sfbay.sun.com/~ohair/webrevs/jdk7/nonfcs-version/webrev/ Bit hard to review if the host isn't reachable :) Could you just attach the patch to your email, or post it on some publicly reachable machine? Thanks, Mark
Re: please disregard my previous email about jaxp 1.4.5 -> Re: jaxp 1.4.5 -> Re: Need reviewer: jaxp and jaxws urls in openjdk6 and openjdk7
On Wed, 2010-11-24 at 10:50 -0800, Joe Wang wrote: > Sorry Mark, and all on build-dev. I didn't pay attention to the > build-dev, I mean didn't realize the original email was sent to an > outside alias. It's my fault. My notice was ONLY a heads-up for Kelly. I > shouldn't have hit reply to all. Please disregard my previous email. I do think these are good things to discuss on the public list. If we want to coordinate on updated jaxp for openjdk7 then we should have that discussion on one of the open lists. So please do post with your ideas and intentions. Thanks, Mark
Need reviewer: NONFCS_BUILD_INFO to add to the non-fcs version string
I need a reviewer for this change: 6987107: Add NONFCS_BUILD_INFO variable to add to but not modify MILESTONE in version string http://javaweb.sfbay.sun.com/~ohair/webrevs/jdk7/nonfcs-version/webrev/ It removed use of JPRT_BUILD_VERSION, adds a more generic NONFCS_BUILD_INFO. Also optimized a few $(shell) uses. -kto
please disregard my previous email about jaxp 1.4.5 -> Re: jaxp 1.4.5 -> Re: Need reviewer: jaxp and jaxws urls in openjdk6 and openjdk7
Sorry Mark, and all on build-dev. I didn't pay attention to the build-dev, I mean didn't realize the original email was sent to an outside alias. It's my fault. My notice was ONLY a heads-up for Kelly. I shouldn't have hit reply to all. Please disregard my previous email. Thanks, Joe On 11/24/2010 10:43 AM, Mark Wielaard wrote: On Wed, 2010-11-24 at d10:20 -0800, Joe Wang wrote: On 11/24/2010 10:13 AM, Kelly O'Hair wrote: On Nov 24, 2010, at 10:10 AM, Joe Wang wrote: By the way, we just finalized a plan on jaxp 1.4.5. We are planning make an internal release first by FF (12/15) of JDK7, to give you so me time to prepare for the integration, we've set 12/8 as the date by which I will provide a source bundle. After that, I will continue working on some bug fixes until 2/2011. The plan is to release jaxp 1.4.5 in 2/2011 and have another integration. A "detail" link to jaxp 1.4.5 is here: http://wiki.se.oracle.com/display/JPG/JAXP+1.4.5 Ok. I assume we need some kind of approval to add this to jdk7? Might want to get that ball rolling if it isn't already. Or any CCC request that might be needed. It's already in the JDK7 PRD [1]. Also, there is no API changes, just some Apache update and bug fixes required by the exit criteria. [1] http://wiki.se.oracle.com/display/JPG/JDK+7+Engineering+Feature+Plans +Overview There is something wrong with these URLs. Neither of them work. It is as if the wiki.se.oracle.com site doesn't exist. Could this be moved to a public location so we can all discuss it? Thanks, Mark
Re: jaxp 1.4.5 -> Re: Need reviewer: jaxp and jaxws urls in openjdk6 and openjdk7
On Wed, 2010-11-24 at 10:20 -0800, Joe Wang wrote: > On 11/24/2010 10:13 AM, Kelly O'Hair wrote: > > On Nov 24, 2010, at 10:10 AM, Joe Wang wrote: > >> By the way, we just finalized a plan on jaxp 1.4.5. We are planning > >> make an internal release first by FF (12/15) of JDK7, to give you so > >> me time to prepare for the integration, we've set 12/8 as the date by > >> which I will provide a source bundle. After that, I will continue > >> working on some bug fixes until 2/2011. The plan is to release jaxp > >> 1.4.5 in 2/2011 and have another integration. A "detail" link to jaxp > >> 1.4.5 is here: http://wiki.se.oracle.com/display/JPG/JAXP+1.4.5 > > > > Ok. I assume we need some kind of approval to add this to jdk7? > > Might want to get that ball rolling if it isn't already. > > Or any CCC request that might be needed. > It's already in the JDK7 PRD [1]. Also, there is no API changes, just > some Apache update and bug fixes required by the exit criteria. > > [1] > http://wiki.se.oracle.com/display/JPG/JDK+7+Engineering+Feature+Plans +Overview There is something wrong with these URLs. Neither of them work. It is as if the wiki.se.oracle.com site doesn't exist. Could this be moved to a public location so we can all discuss it? Thanks, Mark
Re: jaxp 1.4.5 -> Re: Need reviewer: jaxp and jaxws urls in openjdk6 and openjdk7
It's already in the JDK7 PRD [1]. Also, there is no API changes, just some Apache update and bug fixes required by the exit criteria. [1] http://wiki.se.oracle.com/display/JPG/JDK+7+Engineering+Feature+Plans+Overview --Joe On 11/24/2010 10:13 AM, Kelly O'Hair wrote: Ok. I assume we need some kind of approval to add this to jdk7? Might want to get that ball rolling if it isn't already. Or any CCC request that might be needed. -kto On Nov 24, 2010, at 10:10 AM, Joe Wang wrote: Kelly, By the way, we just finalized a plan on jaxp 1.4.5. We are planning make an internal release first by FF (12/15) of JDK7, to give you so me time to prepare for the integration, we've set 12/8 as the date by which I will provide a source bundle. After that, I will continue working on some bug fixes until 2/2011. The plan is to release jaxp 1.4.5 in 2/2011 and have another integration. A "detail" link to jaxp 1.4.5 is here: http://wiki.se.oracle.com/display/JPG/JAXP+1.4.5 Thanks, Joe On 11/23/2010 9:20 AM, Kelly O'Hair wrote: Need reviewer: 7002248: Update urls for jaxp and jaxws source downloads Changes to use the these download areas: https://java.net/downloads/jaxp/jdk7 https://java.net/downloads/jax-ws/JDK7/ https://java.net/downloads/jax-ws/OpenJDK6/ Webrevs: http://cr.openjdk.java.net/~ohair/openjdk6/url-change/webrev/ http://cr.openjdk.java.net/~ohair/openjdk7/url-change/webrev/ The file jaxp-1_4_4.zip only seems to be in the jdk7 download area, so both openjdk6 and openjdk7 refer to that same bundle. -kto
Re: jaxp 1.4.5 -> Re: Need reviewer: jaxp and jaxws urls in openjdk6 and openjdk7
Ok. I assume we need some kind of approval to add this to jdk7? Might want to get that ball rolling if it isn't already. Or any CCC request that might be needed. -kto On Nov 24, 2010, at 10:10 AM, Joe Wang wrote: Kelly, By the way, we just finalized a plan on jaxp 1.4.5. We are planning make an internal release first by FF (12/15) of JDK7, to give you so me time to prepare for the integration, we've set 12/8 as the date by which I will provide a source bundle. After that, I will continue working on some bug fixes until 2/2011. The plan is to release jaxp 1.4.5 in 2/2011 and have another integration. A "detail" link to jaxp 1.4.5 is here: http://wiki.se.oracle.com/display/JPG/JAXP+1.4.5 Thanks, Joe On 11/23/2010 9:20 AM, Kelly O'Hair wrote: Need reviewer: 7002248: Update urls for jaxp and jaxws source downloads Changes to use the these download areas: https://java.net/downloads/jaxp/jdk7 https://java.net/downloads/jax-ws/JDK7/ https://java.net/downloads/jax-ws/OpenJDK6/ Webrevs: http://cr.openjdk.java.net/~ohair/openjdk6/url-change/webrev/ http://cr.openjdk.java.net/~ohair/openjdk7/url-change/webrev/ The file jaxp-1_4_4.zip only seems to be in the jdk7 download area, so both openjdk6 and openjdk7 refer to that same bundle. -kto
jaxp 1.4.5 -> Re: Need reviewer: jaxp and jaxws urls in openjdk6 and openjdk7
Kelly, By the way, we just finalized a plan on jaxp 1.4.5. We are planning make an internal release first by FF (12/15) of JDK7, to give you so me time to prepare for the integration, we've set 12/8 as the date by which I will provide a source bundle. After that, I will continue working on some bug fixes until 2/2011. The plan is to release jaxp 1.4.5 in 2/2011 and have another integration. A "detail" link to jaxp 1.4.5 is here: http://wiki.se.oracle.com/display/JPG/JAXP+1.4.5 Thanks, Joe On 11/23/2010 9:20 AM, Kelly O'Hair wrote: Need reviewer: 7002248: Update urls for jaxp and jaxws source downloads Changes to use the these download areas: https://java.net/downloads/jaxp/jdk7 https://java.net/downloads/jax-ws/JDK7/ https://java.net/downloads/jax-ws/OpenJDK6/ Webrevs: http://cr.openjdk.java.net/~ohair/openjdk6/url-change/webrev/ http://cr.openjdk.java.net/~ohair/openjdk7/url-change/webrev/ The file jaxp-1_4_4.zip only seems to be in the jdk7 download area, so both openjdk6 and openjdk7 refer to that same bundle. -kto
Re: Need reviewer: jaxp and jaxws urls in openjdk6 and openjdk7
I can confirm the URLs for jaxp are correct. Yes, the file jaxp-1_4_4.zip and jaxp-unittests-1_4_4.zip only exist in the jdk7 download area, so both openjdk6 and openjdk7 refer to that same bundles. By the way, I'm asking around about using download.java.net/jaxp to hold jaxp artifacts in the future. It seems to me that's the same download server as what Kelly pointed out earlier for openjdk, e.g. http://download.java.net/openjdk/jdk7/ --Joe On 11/23/2010 9:33 AM, Joe Darcy wrote: Both sets of changes approved; thanks Kelly. Joe (Wang) and Rama, please verify the new download areas are also populated with the older source bundles so that the earlier builds can be recreated. -Joe Kelly O'Hair wrote: Need reviewer: 7002248: Update urls for jaxp and jaxws source downloads Changes to use the these download areas: https://java.net/downloads/jaxp/jdk7 https://java.net/downloads/jax-ws/JDK7/ https://java.net/downloads/jax-ws/OpenJDK6/ Webrevs: http://cr.openjdk.java.net/~ohair/openjdk6/url-change/webrev/ http://cr.openjdk.java.net/~ohair/openjdk7/url-change/webrev/ The file jaxp-1_4_4.zip only seems to be in the jdk7 download area, so both openjdk6 and openjdk7 refer to that same bundle. -kto