Re: Java SCA - Including fix numbers in release docs
Just adding a little caveat here, that, only Resolved jiras can be bulk updated later on (when a release is approaching), jiras on the closed as fixed state need to be reopened in order to get edited. If we are going to use bulk updates, we should make everybody resolve the jiras, instead of closing them. PS.: If someone knows a way to bulk edit a closed jira, please let me know. On 5/23/07, kelvin goodson [EMAIL PROTECTED] wrote: I really don't like making unnecessary 'rules' or policy or trying to restrict or control who can do something I think you are probably right, that's why I suggested the alternative rule of thumb, which is nothing more than common sense really. I think the important thing is instilling an understanding that the field will be used at release time and therefore its helpful to do the right thing with it. If we can't find a less relaxed way to do this then I'd prefer to just not include the JIRA list in the release notes. Couldn't it just be whoever adds the jira list to the release notes checks the list is correct and that will also be validated during everyones review of the release? Sure, but in the absence of certainty about the fix-release field accuracy the query that the RM must use would probably be based on the fixes that occurred between the revision numbers for when the branches and tags of the previous and current releases were cut, taking into account any porting of fixes between trunk/branch/tag after their creation. It works, I'm sure, it's just harder work :-( Kelvin. ...ant On 5/22/07, Luciano Resende [EMAIL PROTECTED] wrote: I think using XXX-Next seems more appropriate now, that we are going out of milestone releases. As for the JIRA process, I think that Kevin's original proposal seems good and would be consistent no matter witch phase of development/release we are, it also leaves room to the Release Manager to control the open issues, like Ant suggested, as the RM can start moving open issues to a specific fix version when approaching the release time. As for Release process, some info available at [1] and we could probably make it more generic to be a Tuscany release process. [1] http://cwiki.apache.org/confluence/display/TUSCANY/Release+Process On 5/22/07, kelvin goodson [EMAIL PROTECTED] wrote: I think my proposal is consistent with your desire to get the overview. When entering the new release phase, all JIRAs fixed in the period since the last release would be reclassified to the newly created version tag, along with all JIRAs that the community sees as important for the forthcoming release. However, an alternative rule of thumb would be that its always safe to use the *Next version as the fix version, whether raising or resolving a JIRA. Only use a specific version if you really are sure that either the resolution of the defect is a blocker for a release or that the fix you have committed will definitely make it into a release. I just liked the simplicity of my original proposal. Kelvin On 22/05/07, ant elder [EMAIL PROTECTED] wrote: One of the problems with not assigning the specific fix version to JIRA's till the end is that you can't see whats outstanding from the JIRA overview page which is something I've found useful and have used it in past releases to manage what things need to get done. See http://issues.apache.org/jira/browse/TUSCANY Maybe just more knowledge about how the versions get used would be enough? ...ant On 5/22/07, kelvin goodson [EMAIL PROTECTED] wrote: Java SDO has been doing this using an Java-SDO-Mx release rather than Java-SDO-Next, but as I said on IRC I think the Next naming is much better. I propose that we adopt the policy that no-one other than a release manager ever assigns anything other than a *Next value for the fix release of a JIRA. The reason I say this is that it makes it simpler around the time of the release. I noted that at the time of the recent SDO release a couple of JIRAs got closed with a fix-version of beta1 after the last release candidate had been cut, but before the beta1 had been released. As there is this time of uncertainty I think its far better to leave the job of assigning a real fix-release value to a JIRA. Its easy for the RM to do a bulk change on all *Next jiras at the appropriate time to whatever the real release becomes know as. Regards, Kelvin. On 21/05/07, haleh mahbod [EMAIL PROTECTED] wrote: It would be good if all subprojects used whatever scheme it is agreed to so a developer going across projects does not have to think about adjusting. On 5/21/07, Simon Laws [EMAIL
Re: Java SCA - Including fix numbers in release docs
Yes +1 to XXX-Next. I really don't like making unnecessary 'rules' or policy or trying to restrict or control who can do something. If we can't find a less relaxed way to do this then I'd prefer to just not include the JIRA list in the release notes. Couldn't it just be whoever adds the jira list to the release notes checks the list is correct and that will also be validated during everyones review of the release? ...ant On 5/22/07, Luciano Resende [EMAIL PROTECTED] wrote: I think using XXX-Next seems more appropriate now, that we are going out of milestone releases. As for the JIRA process, I think that Kevin's original proposal seems good and would be consistent no matter witch phase of development/release we are, it also leaves room to the Release Manager to control the open issues, like Ant suggested, as the RM can start moving open issues to a specific fix version when approaching the release time. As for Release process, some info available at [1] and we could probably make it more generic to be a Tuscany release process. [1] http://cwiki.apache.org/confluence/display/TUSCANY/Release+Process On 5/22/07, kelvin goodson [EMAIL PROTECTED] wrote: I think my proposal is consistent with your desire to get the overview. When entering the new release phase, all JIRAs fixed in the period since the last release would be reclassified to the newly created version tag, along with all JIRAs that the community sees as important for the forthcoming release. However, an alternative rule of thumb would be that its always safe to use the *Next version as the fix version, whether raising or resolving a JIRA. Only use a specific version if you really are sure that either the resolution of the defect is a blocker for a release or that the fix you have committed will definitely make it into a release. I just liked the simplicity of my original proposal. Kelvin On 22/05/07, ant elder [EMAIL PROTECTED] wrote: One of the problems with not assigning the specific fix version to JIRA's till the end is that you can't see whats outstanding from the JIRA overview page which is something I've found useful and have used it in past releases to manage what things need to get done. See http://issues.apache.org/jira/browse/TUSCANY Maybe just more knowledge about how the versions get used would be enough? ...ant On 5/22/07, kelvin goodson [EMAIL PROTECTED] wrote: Java SDO has been doing this using an Java-SDO-Mx release rather than Java-SDO-Next, but as I said on IRC I think the Next naming is much better. I propose that we adopt the policy that no-one other than a release manager ever assigns anything other than a *Next value for the fix release of a JIRA. The reason I say this is that it makes it simpler around the time of the release. I noted that at the time of the recent SDO release a couple of JIRAs got closed with a fix-version of beta1 after the last release candidate had been cut, but before the beta1 had been released. As there is this time of uncertainty I think its far better to leave the job of assigning a real fix-release value to a JIRA. Its easy for the RM to do a bulk change on all *Next jiras at the appropriate time to whatever the real release becomes know as. Regards, Kelvin. On 21/05/07, haleh mahbod [EMAIL PROTECTED] wrote: It would be good if all subprojects used whatever scheme it is agreed to so a developer going across projects does not have to think about adjusting. On 5/21/07, Simon Laws [EMAIL PROTECTED] wrote: This time round, as so much had changed, we didn't include JIRA numbers in the release docs. It seems like a good thing to do in the future though. If everyone agrees that this is a good thing we need to be fairly organized about how we use JIRA otherwise we suffer a lot of pain come release time working out what the list should look like. So, from the IRC today, it has been suggested that we take care to note what release a fix targets using the protocol that the release is Java-SCA-Next until we get to release time and decide what the release number is. At that point we switch over all the fixes that make the release to the right number. This may well have been the intention all along as I note that the Java-SCA-Next category has a lot of fixes in it. I'll take a look through it and see if I can work out what the state of play is so we can start filling it up again. Anything else we should be doing with respect to JIRA to make the release process easier? Simon -- Luciano Resende Apache Tuscany Committer http://people.apache.org/~lresende http://lresende.blogspot.com/
Re: Java SCA - Including fix numbers in release docs
+1 XXX-Next I don't mind who assigns JIRA to the release number. It can't be done though until we know that the release number will be (was quite late in the last cycle). Hopefully we will do more frequent releases from now so there will be few JIRAs in the XXX-Next box and more with the actual release tag. Simon
Re: Java SCA - Including fix numbers in release docs
I really don't like making unnecessary 'rules' or policy or trying to restrict or control who can do something I think you are probably right, that's why I suggested the alternative rule of thumb, which is nothing more than common sense really. I think the important thing is instilling an understanding that the field will be used at release time and therefore its helpful to do the right thing with it. If we can't find a less relaxed way to do this then I'd prefer to just not include the JIRA list in the release notes. Couldn't it just be whoever adds the jira list to the release notes checks the list is correct and that will also be validated during everyones review of the release? Sure, but in the absence of certainty about the fix-release field accuracy the query that the RM must use would probably be based on the fixes that occurred between the revision numbers for when the branches and tags of the previous and current releases were cut, taking into account any porting of fixes between trunk/branch/tag after their creation. It works, I'm sure, it's just harder work :-( Kelvin. ...ant On 5/22/07, Luciano Resende [EMAIL PROTECTED] wrote: I think using XXX-Next seems more appropriate now, that we are going out of milestone releases. As for the JIRA process, I think that Kevin's original proposal seems good and would be consistent no matter witch phase of development/release we are, it also leaves room to the Release Manager to control the open issues, like Ant suggested, as the RM can start moving open issues to a specific fix version when approaching the release time. As for Release process, some info available at [1] and we could probably make it more generic to be a Tuscany release process. [1] http://cwiki.apache.org/confluence/display/TUSCANY/Release+Process On 5/22/07, kelvin goodson [EMAIL PROTECTED] wrote: I think my proposal is consistent with your desire to get the overview. When entering the new release phase, all JIRAs fixed in the period since the last release would be reclassified to the newly created version tag, along with all JIRAs that the community sees as important for the forthcoming release. However, an alternative rule of thumb would be that its always safe to use the *Next version as the fix version, whether raising or resolving a JIRA. Only use a specific version if you really are sure that either the resolution of the defect is a blocker for a release or that the fix you have committed will definitely make it into a release. I just liked the simplicity of my original proposal. Kelvin On 22/05/07, ant elder [EMAIL PROTECTED] wrote: One of the problems with not assigning the specific fix version to JIRA's till the end is that you can't see whats outstanding from the JIRA overview page which is something I've found useful and have used it in past releases to manage what things need to get done. See http://issues.apache.org/jira/browse/TUSCANY Maybe just more knowledge about how the versions get used would be enough? ...ant On 5/22/07, kelvin goodson [EMAIL PROTECTED] wrote: Java SDO has been doing this using an Java-SDO-Mx release rather than Java-SDO-Next, but as I said on IRC I think the Next naming is much better. I propose that we adopt the policy that no-one other than a release manager ever assigns anything other than a *Next value for the fix release of a JIRA. The reason I say this is that it makes it simpler around the time of the release. I noted that at the time of the recent SDO release a couple of JIRAs got closed with a fix-version of beta1 after the last release candidate had been cut, but before the beta1 had been released. As there is this time of uncertainty I think its far better to leave the job of assigning a real fix-release value to a JIRA. Its easy for the RM to do a bulk change on all *Next jiras at the appropriate time to whatever the real release becomes know as. Regards, Kelvin. On 21/05/07, haleh mahbod [EMAIL PROTECTED] wrote: It would be good if all subprojects used whatever scheme it is agreed to so a developer going across projects does not have to think about adjusting. On 5/21/07, Simon Laws [EMAIL PROTECTED] wrote: This time round, as so much had changed, we didn't include JIRA numbers in the release docs. It seems like a good thing to do in the future though. If everyone agrees that this is a good thing we need to be fairly organized about how we use JIRA otherwise we suffer a lot of pain come release time working out what the list should look like. So, from the IRC today, it has been suggested that we take care to note what release a fix targets
Re: Java SCA - Including fix numbers in release docs
OK, I agree. I note that SDO already does this. Any more thoughts about things we can do to improve the way we can all work in this area? What else do SDO, DAS do that SCA needs to do? Maybe we can come up with a checklist for the build and release up on the developer guide section of the web site. I know Kelvin started collecting this information for SDO but can't lay my finger on it directly. Simon
Re: Java SCA - Including fix numbers in release docs
Java SDO has been doing this using an Java-SDO-Mx release rather than Java-SDO-Next, but as I said on IRC I think the Next naming is much better. I propose that we adopt the policy that no-one other than a release manager ever assigns anything other than a *Next value for the fix release of a JIRA. The reason I say this is that it makes it simpler around the time of the release. I noted that at the time of the recent SDO release a couple of JIRAs got closed with a fix-version of beta1 after the last release candidate had been cut, but before the beta1 had been released. As there is this time of uncertainty I think its far better to leave the job of assigning a real fix-release value to a JIRA. Its easy for the RM to do a bulk change on all *Next jiras at the appropriate time to whatever the real release becomes know as. Regards, Kelvin. On 21/05/07, haleh mahbod [EMAIL PROTECTED] wrote: It would be good if all subprojects used whatever scheme it is agreed to so a developer going across projects does not have to think about adjusting. On 5/21/07, Simon Laws [EMAIL PROTECTED] wrote: This time round, as so much had changed, we didn't include JIRA numbers in the release docs. It seems like a good thing to do in the future though. If everyone agrees that this is a good thing we need to be fairly organized about how we use JIRA otherwise we suffer a lot of pain come release time working out what the list should look like. So, from the IRC today, it has been suggested that we take care to note what release a fix targets using the protocol that the release is Java-SCA-Next until we get to release time and decide what the release number is. At that point we switch over all the fixes that make the release to the right number. This may well have been the intention all along as I note that the Java-SCA-Next category has a lot of fixes in it. I'll take a look through it and see if I can work out what the state of play is so we can start filling it up again. Anything else we should be doing with respect to JIRA to make the release process easier? Simon
Re: Java SCA - Including fix numbers in release docs
One of the problems with not assigning the specific fix version to JIRA's till the end is that you can't see whats outstanding from the JIRA overview page which is something I've found useful and have used it in past releases to manage what things need to get done. See http://issues.apache.org/jira/browse/TUSCANY Maybe just more knowledge about how the versions get used would be enough? ...ant On 5/22/07, kelvin goodson [EMAIL PROTECTED] wrote: Java SDO has been doing this using an Java-SDO-Mx release rather than Java-SDO-Next, but as I said on IRC I think the Next naming is much better. I propose that we adopt the policy that no-one other than a release manager ever assigns anything other than a *Next value for the fix release of a JIRA. The reason I say this is that it makes it simpler around the time of the release. I noted that at the time of the recent SDO release a couple of JIRAs got closed with a fix-version of beta1 after the last release candidate had been cut, but before the beta1 had been released. As there is this time of uncertainty I think its far better to leave the job of assigning a real fix-release value to a JIRA. Its easy for the RM to do a bulk change on all *Next jiras at the appropriate time to whatever the real release becomes know as. Regards, Kelvin. On 21/05/07, haleh mahbod [EMAIL PROTECTED] wrote: It would be good if all subprojects used whatever scheme it is agreed to so a developer going across projects does not have to think about adjusting. On 5/21/07, Simon Laws [EMAIL PROTECTED] wrote: This time round, as so much had changed, we didn't include JIRA numbers in the release docs. It seems like a good thing to do in the future though. If everyone agrees that this is a good thing we need to be fairly organized about how we use JIRA otherwise we suffer a lot of pain come release time working out what the list should look like. So, from the IRC today, it has been suggested that we take care to note what release a fix targets using the protocol that the release is Java-SCA-Next until we get to release time and decide what the release number is. At that point we switch over all the fixes that make the release to the right number. This may well have been the intention all along as I note that the Java-SCA-Next category has a lot of fixes in it. I'll take a look through it and see if I can work out what the state of play is so we can start filling it up again. Anything else we should be doing with respect to JIRA to make the release process easier? Simon
Re: Java SCA - Including fix numbers in release docs
I think my proposal is consistent with your desire to get the overview. When entering the new release phase, all JIRAs fixed in the period since the last release would be reclassified to the newly created version tag, along with all JIRAs that the community sees as important for the forthcoming release. However, an alternative rule of thumb would be that its always safe to use the *Next version as the fix version, whether raising or resolving a JIRA. Only use a specific version if you really are sure that either the resolution of the defect is a blocker for a release or that the fix you have committed will definitely make it into a release. I just liked the simplicity of my original proposal. Kelvin On 22/05/07, ant elder [EMAIL PROTECTED] wrote: One of the problems with not assigning the specific fix version to JIRA's till the end is that you can't see whats outstanding from the JIRA overview page which is something I've found useful and have used it in past releases to manage what things need to get done. See http://issues.apache.org/jira/browse/TUSCANY Maybe just more knowledge about how the versions get used would be enough? ...ant On 5/22/07, kelvin goodson [EMAIL PROTECTED] wrote: Java SDO has been doing this using an Java-SDO-Mx release rather than Java-SDO-Next, but as I said on IRC I think the Next naming is much better. I propose that we adopt the policy that no-one other than a release manager ever assigns anything other than a *Next value for the fix release of a JIRA. The reason I say this is that it makes it simpler around the time of the release. I noted that at the time of the recent SDO release a couple of JIRAs got closed with a fix-version of beta1 after the last release candidate had been cut, but before the beta1 had been released. As there is this time of uncertainty I think its far better to leave the job of assigning a real fix-release value to a JIRA. Its easy for the RM to do a bulk change on all *Next jiras at the appropriate time to whatever the real release becomes know as. Regards, Kelvin. On 21/05/07, haleh mahbod [EMAIL PROTECTED] wrote: It would be good if all subprojects used whatever scheme it is agreed to so a developer going across projects does not have to think about adjusting. On 5/21/07, Simon Laws [EMAIL PROTECTED] wrote: This time round, as so much had changed, we didn't include JIRA numbers in the release docs. It seems like a good thing to do in the future though. If everyone agrees that this is a good thing we need to be fairly organized about how we use JIRA otherwise we suffer a lot of pain come release time working out what the list should look like. So, from the IRC today, it has been suggested that we take care to note what release a fix targets using the protocol that the release is Java-SCA-Next until we get to release time and decide what the release number is. At that point we switch over all the fixes that make the release to the right number. This may well have been the intention all along as I note that the Java-SCA-Next category has a lot of fixes in it. I'll take a look through it and see if I can work out what the state of play is so we can start filling it up again. Anything else we should be doing with respect to JIRA to make the release process easier? Simon
Re: Java SCA - Including fix numbers in release docs
I think using XXX-Next seems more appropriate now, that we are going out of milestone releases. As for the JIRA process, I think that Kevin's original proposal seems good and would be consistent no matter witch phase of development/release we are, it also leaves room to the Release Manager to control the open issues, like Ant suggested, as the RM can start moving open issues to a specific fix version when approaching the release time. As for Release process, some info available at [1] and we could probably make it more generic to be a Tuscany release process. [1] http://cwiki.apache.org/confluence/display/TUSCANY/Release+Process On 5/22/07, kelvin goodson [EMAIL PROTECTED] wrote: I think my proposal is consistent with your desire to get the overview. When entering the new release phase, all JIRAs fixed in the period since the last release would be reclassified to the newly created version tag, along with all JIRAs that the community sees as important for the forthcoming release. However, an alternative rule of thumb would be that its always safe to use the *Next version as the fix version, whether raising or resolving a JIRA. Only use a specific version if you really are sure that either the resolution of the defect is a blocker for a release or that the fix you have committed will definitely make it into a release. I just liked the simplicity of my original proposal. Kelvin On 22/05/07, ant elder [EMAIL PROTECTED] wrote: One of the problems with not assigning the specific fix version to JIRA's till the end is that you can't see whats outstanding from the JIRA overview page which is something I've found useful and have used it in past releases to manage what things need to get done. See http://issues.apache.org/jira/browse/TUSCANY Maybe just more knowledge about how the versions get used would be enough? ...ant On 5/22/07, kelvin goodson [EMAIL PROTECTED] wrote: Java SDO has been doing this using an Java-SDO-Mx release rather than Java-SDO-Next, but as I said on IRC I think the Next naming is much better. I propose that we adopt the policy that no-one other than a release manager ever assigns anything other than a *Next value for the fix release of a JIRA. The reason I say this is that it makes it simpler around the time of the release. I noted that at the time of the recent SDO release a couple of JIRAs got closed with a fix-version of beta1 after the last release candidate had been cut, but before the beta1 had been released. As there is this time of uncertainty I think its far better to leave the job of assigning a real fix-release value to a JIRA. Its easy for the RM to do a bulk change on all *Next jiras at the appropriate time to whatever the real release becomes know as. Regards, Kelvin. On 21/05/07, haleh mahbod [EMAIL PROTECTED] wrote: It would be good if all subprojects used whatever scheme it is agreed to so a developer going across projects does not have to think about adjusting. On 5/21/07, Simon Laws [EMAIL PROTECTED] wrote: This time round, as so much had changed, we didn't include JIRA numbers in the release docs. It seems like a good thing to do in the future though. If everyone agrees that this is a good thing we need to be fairly organized about how we use JIRA otherwise we suffer a lot of pain come release time working out what the list should look like. So, from the IRC today, it has been suggested that we take care to note what release a fix targets using the protocol that the release is Java-SCA-Next until we get to release time and decide what the release number is. At that point we switch over all the fixes that make the release to the right number. This may well have been the intention all along as I note that the Java-SCA-Next category has a lot of fixes in it. I'll take a look through it and see if I can work out what the state of play is so we can start filling it up again. Anything else we should be doing with respect to JIRA to make the release process easier? Simon -- Luciano Resende Apache Tuscany Committer http://people.apache.org/~lresende http://lresende.blogspot.com/
Re: Java SCA - Including fix numbers in release docs
It would be good if all subprojects used whatever scheme it is agreed to so a developer going across projects does not have to think about adjusting. On 5/21/07, Simon Laws [EMAIL PROTECTED] wrote: This time round, as so much had changed, we didn't include JIRA numbers in the release docs. It seems like a good thing to do in the future though. If everyone agrees that this is a good thing we need to be fairly organized about how we use JIRA otherwise we suffer a lot of pain come release time working out what the list should look like. So, from the IRC today, it has been suggested that we take care to note what release a fix targets using the protocol that the release is Java-SCA-Next until we get to release time and decide what the release number is. At that point we switch over all the fixes that make the release to the right number. This may well have been the intention all along as I note that the Java-SCA-Next category has a lot of fixes in it. I'll take a look through it and see if I can work out what the state of play is so we can start filling it up again. Anything else we should be doing with respect to JIRA to make the release process easier? Simon
Re: Java SCA - Including fix numbers in release docs
I agree with Haleh, we should try to have consistence on areas common to all Tuscany sub-projects, JIRA, Distributions, Release Process, etc On 5/21/07, haleh mahbod [EMAIL PROTECTED] wrote: It would be good if all subprojects used whatever scheme it is agreed to so a developer going across projects does not have to think about adjusting. On 5/21/07, Simon Laws [EMAIL PROTECTED] wrote: This time round, as so much had changed, we didn't include JIRA numbers in the release docs. It seems like a good thing to do in the future though. If everyone agrees that this is a good thing we need to be fairly organized about how we use JIRA otherwise we suffer a lot of pain come release time working out what the list should look like. So, from the IRC today, it has been suggested that we take care to note what release a fix targets using the protocol that the release is Java-SCA-Next until we get to release time and decide what the release number is. At that point we switch over all the fixes that make the release to the right number. This may well have been the intention all along as I note that the Java-SCA-Next category has a lot of fixes in it. I'll take a look through it and see if I can work out what the state of play is so we can start filling it up again. Anything else we should be doing with respect to JIRA to make the release process easier? Simon -- Luciano Resende Apache Tuscany Committer http://people.apache.org/~lresende http://lresende.blogspot.com/