[
https://issues.apache.org/jira/browse/HBASE-19674?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16309444#comment-16309444
]
Jan Hentschel commented on HBASE-19674:
---------------------------------------
My point is, instead of changing the semantics of {{make_patch}} to something
new it should be unified to follow the naming scheme of the {{submit-patch}}
script (and what is described in the book).
> make_patch.sh version increment fails
> -------------------------------------
>
> Key: HBASE-19674
> URL: https://issues.apache.org/jira/browse/HBASE-19674
> Project: HBase
> Issue Type: Improvement
> Reporter: Niels Basjes
> Assignee: Niels Basjes
> Attachments: HBASE-19674.20171230-131310.patch,
> HBASE-19674.20171230-152443.patch
>
>
> I have 5 things in the {{make_patch.sh}} script where I see room for
> improvement:
> 1) BUG:
> Assume my working branch is called {{HBASE-19673}}
> Now if I run
> {{dev-support/make_patch.sh -b origin/branch-1}}
> a patch is created with the name
> {{~/patches/HBASE-19673.v1.branch-1.patch}}
> When I run the same command again the version is not incremented.
> The reason is that the script checks for {{HBASE-19673.v1.patch}} which is
> without the branch name.
> 2) Messy: The first patch created does NOT include the version tag at all.
> 3) Messy: The version starts with '1' so when we reach patch '10' they will
> be ordered incorrectly in Jira (which is based on name)
> 4) New feature: I personally prefer using the timestamp as the 'version' of
> the patch because these are much easier to order.
> 5) Messy: If you for example only have one file {{HBASE-19674.v05.patch}}
> then the next file generated will be {{HBASE-19674.v01.patch}} instead of the
> expected {{HBASE-19674.v06.patch}}
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
