hadoop-common-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Kirby Bohling <kirby.bohl...@gmail.com>
Subject Re: JIRA attachments order
Date Fri, 09 Sep 2011 20:38:57 GMT
On Fri, Sep 9, 2011 at 3:15 PM, Aaron T. Myers <atm@cloudera.com> wrote:
> On Fri, Sep 9, 2011 at 12:57 PM, Eli Collins <eli@cloudera.com> wrote:
>> Patches for a specific branch should be named:  jira-xyz-branch.patch
>> where "branch" may be abbreviated, eg hdfs-123-security.patch
> +1, if we ever hope to implement HADOOP-7435 [1], it will be necessary to
> standardize the branch-name-in-patch-name scheme.

Coming out of lurker mode...

Someday I wish Apache would find/adopt a distributed version control
system (I know about git.apache.org, but I mean pushing that further),
and use something like gerrit or review board.  So if you have a
patch, or a series of patches, you'd just use a version control
system.  Make it such that when you submit to review system you have
to log in and say "Yes, I want contribute this under to the ASF under
an Apache 2.0 license".  Make me use my JIRA/Apache credentials to
submit reviews directly from my version control, and force the commit
message to be properly formatted to tie it to a JIRA issue.  Make the
patch submission have a link right next to it to take me to the review
system, where it can be +1/0/-1'ed with commentary in line.

I mean this as nicely as possible, but generating patches and
uploading them feels like you're stuck in the stone age (also known as
the mid 1990s).  95% of the problems being discussed in this thread
are irrelevant if the tools chosen matched up with the problems.  This
whole discussion feels like discussing the merits of using the butt
end of a screw driver or a monkey wrench to drive nails.  Why not just
pick the nice hammer lying over there (even if it requires some
assembly).  Even if you don't feel like moving completely away from
SVN and JIRA (which I understand the stability of the technical and
legal aspects are important to the heavy lifting Apache does to ensure
everything is on the up and up), if you decided that you would support
git or mercurial (or DVCS of choice), along with a web based code
review integration (what ever one you like), could do the work of
submitting patches and labeling revisions for you.

I think that'd be the path of least resistance, and you'd see
developers migrate rapidly to using that workflow, and this whole
thing is mostly moot.  At least IMHO.

... Back to lurking mode...   Thanks for all the great code.


> --
> Aaron T. Myers
> Software Engineer, Cloudera
> [1] https://issues.apache.org/jira/browse/HADOOP-7435

View raw message