Re: moving Hive to git
I'm not convinced git would be better. Could someone please spell out the advantages? In particular: 1. ... git is more powerful and easy to use (once you go past the learning curve!) -- Lefty On Tue, Sep 16, 2014 at 4:27 PM, Sergey Shelukhin ser...@hortonworks.com wrote: It seems there was consensus that we should move. Any volunteers to do it? I can try to find the details on how HBase migrated. On Fri, Sep 12, 2014 at 7:10 PM, Lefty Leverenz leftylever...@gmail.com wrote: We had a related discussion March 5 - 10: llsmugcwkuryr5tb http://markmail.org/message/llsmugcwkuryr5tb through rq66qe2cpfgw7o5s http://markmail.org/message/rq66qe2cpfgw7o5s. -- Lefty On Fri, Sep 12, 2014 at 9:33 PM, Sergey Shelukhin ser...@hortonworks.com wrote: Hi. Many Apache projects are moving to git from svn; HBase moved recently, and as far as I have heard Hadoop has moved too. Are there any objections to moving Hive to use git too? I wanted to start a preliminary discussion. -- CONFIDENTIALITY NOTICE NOTICE: This message is intended for the use of the individual or entity to which it is addressed and may contain information that is confidential, privileged and exempt from disclosure under applicable law. If the reader of this message is not the intended recipient, you are hereby notified that any printing, copying, dissemination, distribution, disclosure or forwarding of this communication is strictly prohibited. If you have received this communication in error, please contact the sender immediately and delete it from your system. Thank You. -- CONFIDENTIALITY NOTICE NOTICE: This message is intended for the use of the individual or entity to which it is addressed and may contain information that is confidential, privileged and exempt from disclosure under applicable law. If the reader of this message is not the intended recipient, you are hereby notified that any printing, copying, dissemination, distribution, disclosure or forwarding of this communication is strictly prohibited. If you have received this communication in error, please contact the sender immediately and delete it from your system. Thank You.
Re: moving Hive to git
For me, the advantages of git are: 1. Each user's working copy contains the global history of the project. So while I'm disconnected in an airplane; I can look at history and logs, switch between branches, and do merges. 2. It makes it very easy to work on development branches and rebase off of trunk to incorporate other people's changes. Tracking of which commits have been merged in to a branch is much better in git than subversion. 3. I can easily share my development branches with others via github (or Apache if we switch). 4. Tags can be signed with pgp keys and unlike subversion, you can't commit into tags by mistake. Thanks, Owen On Wed, Sep 17, 2014 at 8:47 AM, Lefty Leverenz leftylever...@gmail.com wrote: I'm not convinced git would be better. Could someone please spell out the advantages? In particular: 1. ... git is more powerful and easy to use (once you go past the learning curve!) -- Lefty On Tue, Sep 16, 2014 at 4:27 PM, Sergey Shelukhin ser...@hortonworks.com wrote: It seems there was consensus that we should move. Any volunteers to do it? I can try to find the details on how HBase migrated. On Fri, Sep 12, 2014 at 7:10 PM, Lefty Leverenz leftylever...@gmail.com wrote: We had a related discussion March 5 - 10: llsmugcwkuryr5tb http://markmail.org/message/llsmugcwkuryr5tb through rq66qe2cpfgw7o5s http://markmail.org/message/rq66qe2cpfgw7o5s. -- Lefty On Fri, Sep 12, 2014 at 9:33 PM, Sergey Shelukhin ser...@hortonworks.com wrote: Hi. Many Apache projects are moving to git from svn; HBase moved recently, and as far as I have heard Hadoop has moved too. Are there any objections to moving Hive to use git too? I wanted to start a preliminary discussion. -- CONFIDENTIALITY NOTICE NOTICE: This message is intended for the use of the individual or entity to which it is addressed and may contain information that is confidential, privileged and exempt from disclosure under applicable law. If the reader of this message is not the intended recipient, you are hereby notified that any printing, copying, dissemination, distribution, disclosure or forwarding of this communication is strictly prohibited. If you have received this communication in error, please contact the sender immediately and delete it from your system. Thank You. -- CONFIDENTIALITY NOTICE NOTICE: This message is intended for the use of the individual or entity to which it is addressed and may contain information that is confidential, privileged and exempt from disclosure under applicable law. If the reader of this message is not the intended recipient, you are hereby notified that any printing, copying, dissemination, distribution, disclosure or forwarding of this communication is strictly prohibited. If you have received this communication in error, please contact the sender immediately and delete it from your system. Thank You.
Re: moving Hive to git
Can we look at how this is done in Hbase or Hadoop. If these projects migrated on git then I am sure that they might have faced similar issues. On 17 Sep 2014 22:37, E.L. Leverenz e.l.lever...@att.net wrote: This is the rest of the message I meant to send on the moving Hive to git thread, but then did an accidental send. Apache rejected several attempts as spam, so I'm sending this from a different email account. This list summarizes the previous discussion, with my questions/comments: 1. ... git is more powerful and easy to use (once you go past the learning curve!) [Thejas] -- that learning curve still intimidates me, which suggests it might also be daunting for newcomers. 2. Switching to git from svn seems to be a proposal slightly different from that of switching to pull request from the head of the thread. Personally I'm +1 to git, but I think patches are very portable and widely adopted in Hadoop ecosystem and we should keep the practice. [Xuefu] -- could someone explain the patch issue? 3. We need to keep patches in Jira ... having a patch in the jira is critical I feel. We must at least have a perma link to the changes. [Edward] -- again, how are patches different in git? 4. In my read of the Apache git - github integration blog post we cannot use pull requests as patches. Just that we'll be notified of them and could perhaps use them as code review. [Brock] -- okay, perhaps this answers my patch question. 5. One additional item I think we should investigate is disabling merge commits on trunk and feature branches. -- uh oh, I'm slipping backwards on the learning curve. 6. I do not think we want Pull Requests coming at us. Better way is let someone open a git branch for the changes, then we review and merge the branch. [Edward] -- okay, creeping back up the learning curve. 7. I'm +1 on switching to git, but only if we can find a way to disable merge commits to trunk and feature branches. I'm -1 on switching to Github since, as far as I know, it only supports merge based workflows. [Carl] 8. Agree with Carl about git merge commits, they make the changes hard to follow. But it should be OK, if there is no way to disable it in the main git repo, it is a small set of active committers, we can make a policy and expect people to follow it. But we should certainly disable 'git push -f' (and anything as distruptive). [Thejas]-- that small set of committers is growing larger all the time. -- Lefty
Re: moving Hive to git
Hi, I am generally +1 on the proposal. I'd strongly want to disable merge commits. They are far too easy to accidently push. If there is no option to disable them, one option would be to do what we did in flume. Basically: 1) Trunk operates as normal. 2) We always have the next release branch open 3) Every commit is committed to trunk and immediately cherry-picked to the release branch. We could use a script to automate this. Brock
Re: moving Hive to git
I can check how HBase operates without merge commits... cherry-picking seems tedious, at least without the script - too easy to forget, and that would arguably be more harmful than a stray merge commit. On Wed, Sep 17, 2014 at 10:47 AM, Brock Noland br...@cloudera.com wrote: Hi, I am generally +1 on the proposal. I'd strongly want to disable merge commits. They are far too easy to accidently push. If there is no option to disable them, one option would be to do what we did in flume. Basically: 1) Trunk operates as normal. 2) We always have the next release branch open 3) Every commit is committed to trunk and immediately cherry-picked to the release branch. We could use a script to automate this. Brock -- CONFIDENTIALITY NOTICE NOTICE: This message is intended for the use of the individual or entity to which it is addressed and may contain information that is confidential, privileged and exempt from disclosure under applicable law. If the reader of this message is not the intended recipient, you are hereby notified that any printing, copying, dissemination, distribution, disclosure or forwarding of this communication is strictly prohibited. If you have received this communication in error, please contact the sender immediately and delete it from your system. Thank You.
Re: moving Hive to git
I'd say learning curve for git is shorter than for svn, especially for newcomers, since git is widely adopted and used (afaik most people use git mirror for development even now, and only use svn to push). svn in my experience makes it hard to do things, as a general principle. I've had to write bash scripts to replace one git command (e.g. git clean), and branching is difficult. On Wed, Sep 17, 2014 at 9:31 AM, E.L. Leverenz e.l.lever...@att.net wrote: This is the rest of the message I meant to send on the moving Hive to git thread, but then did an accidental send. Apache rejected several attempts as spam, so I'm sending this from a different email account. This list summarizes the previous discussion, with my questions/comments: 1. ... git is more powerful and easy to use (once you go past the learning curve!) [Thejas] -- that learning curve still intimidates me, which suggests it might also be daunting for newcomers. 2. Switching to git from svn seems to be a proposal slightly different from that of switching to pull request from the head of the thread. Personally I'm +1 to git, but I think patches are very portable and widely adopted in Hadoop ecosystem and we should keep the practice. [Xuefu] -- could someone explain the patch issue? 3. We need to keep patches in Jira ... having a patch in the jira is critical I feel. We must at least have a perma link to the changes. [Edward] -- again, how are patches different in git? 4. In my read of the Apache git - github integration blog post we cannot use pull requests as patches. Just that we'll be notified of them and could perhaps use them as code review. [Brock] -- okay, perhaps this answers my patch question. 5. One additional item I think we should investigate is disabling merge commits on trunk and feature branches. -- uh oh, I'm slipping backwards on the learning curve. 6. I do not think we want Pull Requests coming at us. Better way is let someone open a git branch for the changes, then we review and merge the branch. [Edward] -- okay, creeping back up the learning curve. 7. I'm +1 on switching to git, but only if we can find a way to disable merge commits to trunk and feature branches. I'm -1 on switching to Github since, as far as I know, it only supports merge based workflows. [Carl] 8. Agree with Carl about git merge commits, they make the changes hard to follow. But it should be OK, if there is no way to disable it in the main git repo, it is a small set of active committers, we can make a policy and expect people to follow it. But we should certainly disable 'git push -f' (and anything as distruptive). [Thejas]-- that small set of committers is growing larger all the time. -- Lefty -- CONFIDENTIALITY NOTICE NOTICE: This message is intended for the use of the individual or entity to which it is addressed and may contain information that is confidential, privileged and exempt from disclosure under applicable law. If the reader of this message is not the intended recipient, you are hereby notified that any printing, copying, dissemination, distribution, disclosure or forwarding of this communication is strictly prohibited. If you have received this communication in error, please contact the sender immediately and delete it from your system. Thank You.
Re: moving Hive to git
Reg. disabling merge commits, if Apache is ok installing git server-side hook scripts, setting up a pre-receive hook could be a possible option: http://stackoverflow.com/questions/2039773/have-remote-git-repository-refuse-merge-commits-on-push On Wed, Sep 17, 2014 at 10:56 AM, Sergey Shelukhin ser...@hortonworks.com wrote: I can check how HBase operates without merge commits... cherry-picking seems tedious, at least without the script - too easy to forget, and that would arguably be more harmful than a stray merge commit. On Wed, Sep 17, 2014 at 10:47 AM, Brock Noland br...@cloudera.com wrote: Hi, I am generally +1 on the proposal. I'd strongly want to disable merge commits. They are far too easy to accidently push. If there is no option to disable them, one option would be to do what we did in flume. Basically: 1) Trunk operates as normal. 2) We always have the next release branch open 3) Every commit is committed to trunk and immediately cherry-picked to the release branch. We could use a script to automate this. Brock -- CONFIDENTIALITY NOTICE NOTICE: This message is intended for the use of the individual or entity to which it is addressed and may contain information that is confidential, privileged and exempt from disclosure under applicable law. If the reader of this message is not the intended recipient, you are hereby notified that any printing, copying, dissemination, distribution, disclosure or forwarding of this communication is strictly prohibited. If you have received this communication in error, please contact the sender immediately and delete it from your system. Thank You.
Re: moving Hive to git
It seems there was consensus that we should move. Any volunteers to do it? I can try to find the details on how HBase migrated. On Fri, Sep 12, 2014 at 7:10 PM, Lefty Leverenz leftylever...@gmail.com wrote: We had a related discussion March 5 - 10: llsmugcwkuryr5tb http://markmail.org/message/llsmugcwkuryr5tb through rq66qe2cpfgw7o5s http://markmail.org/message/rq66qe2cpfgw7o5s. -- Lefty On Fri, Sep 12, 2014 at 9:33 PM, Sergey Shelukhin ser...@hortonworks.com wrote: Hi. Many Apache projects are moving to git from svn; HBase moved recently, and as far as I have heard Hadoop has moved too. Are there any objections to moving Hive to use git too? I wanted to start a preliminary discussion. -- CONFIDENTIALITY NOTICE NOTICE: This message is intended for the use of the individual or entity to which it is addressed and may contain information that is confidential, privileged and exempt from disclosure under applicable law. If the reader of this message is not the intended recipient, you are hereby notified that any printing, copying, dissemination, distribution, disclosure or forwarding of this communication is strictly prohibited. If you have received this communication in error, please contact the sender immediately and delete it from your system. Thank You. -- CONFIDENTIALITY NOTICE NOTICE: This message is intended for the use of the individual or entity to which it is addressed and may contain information that is confidential, privileged and exempt from disclosure under applicable law. If the reader of this message is not the intended recipient, you are hereby notified that any printing, copying, dissemination, distribution, disclosure or forwarding of this communication is strictly prohibited. If you have received this communication in error, please contact the sender immediately and delete it from your system. Thank You.
Re: moving Hive to git
We had a related discussion March 5 - 10: llsmugcwkuryr5tb http://markmail.org/message/llsmugcwkuryr5tb through rq66qe2cpfgw7o5s http://markmail.org/message/rq66qe2cpfgw7o5s. -- Lefty On Fri, Sep 12, 2014 at 9:33 PM, Sergey Shelukhin ser...@hortonworks.com wrote: Hi. Many Apache projects are moving to git from svn; HBase moved recently, and as far as I have heard Hadoop has moved too. Are there any objections to moving Hive to use git too? I wanted to start a preliminary discussion. -- CONFIDENTIALITY NOTICE NOTICE: This message is intended for the use of the individual or entity to which it is addressed and may contain information that is confidential, privileged and exempt from disclosure under applicable law. If the reader of this message is not the intended recipient, you are hereby notified that any printing, copying, dissemination, distribution, disclosure or forwarding of this communication is strictly prohibited. If you have received this communication in error, please contact the sender immediately and delete it from your system. Thank You.
