Thanks Greg!

On Mon, Apr 28, 2014 at 12:06 PM, Gregory Chanan <[email protected]>wrote:

> I was trying to match up the git tag vs the released artifact, but didn't
> know a standard way of doing that.  As I mentioned in the original e-mail,
> I wasn't sure if what I did was correct, but it was sufficient for my
> purposes since it helped me find the discrepancy.
>
> Running diff -rq on the untared release artifact and the source as of the
> git tag sounds sufficient.  It sounds like I can get the source as of the
> git tag either via git archive or via maven, since they produce the same
> tree, as you mentioned.  If someone knows a better way, I'd love to hear
> it.
>
> I use diff -rq as well and seems to work fine. But would love to hear if
there is a standard way.


> FWIW I chose "git archive" rather than mvn because we are doing a source
> release and git understands what is source.  Invoking mvn seemed like a
> "bad" idea, because:
> 1) it invokes the build, but we are doing a source release that should be
> independent of the build.
> 2) it will pull in files that are not in the source, e.g. the .iml files
> generated by eclipse or even files generated via the build itself
> I see that the how-to-release page lists using mvn to generate the source
> artifact (
> https://cwiki.apache.org/confluence/display/SENTRY/How+to+Release),
> but for the reasons above it seems like we should replace this with using
> git archive.
>
>
I agree that, we should use git archive instead, as we are only releasing
src artifacts. Unless there was a reason for using maven to generate src
artifacts.


> Thoughts?
> Greg
>
>
>
I think we should update the sentry how to release page. Also having a
voting checklist would help. So that the release manager and the voters
have one way of verifying things.


> On Mon, Apr 28, 2014 at 11:51 AM, Sravya Tirukkovalur
> <[email protected]>wrote:
>
> > On Mon, Apr 28, 2014 at 11:32 AM, Sravya Tirukkovalur
> > <[email protected]>wrote:
> >
> > >
> > >
> > >
> > > On Mon, Apr 28, 2014 at 9:15 AM, karthik ramachandran <
> > > [email protected]> wrote:
> > >
> > >> All,
> > >>
> > >> To fix the discrepancies between what is in the RC and the tag, I
> > pulled a
> > >> fresh version of code from git and ran:
> > >>
> > >> git checkout release-1.3.0-rc2
> > >> mvn package -DskipTests
> > >>
> > >> I then re-signed and created new hashes for the resulting gzip and
> > >> uploaded
> > >> those artifacts to
> > >> http://people.apache.org/~kramachandran/sentry-1.3.0-rc2/
> > >>
> > >> When I run diff -r incubator-sentry/
> apache-sentry-1.3.0-incubating-src/
> > >> the diff comes back clean. In this case incubator-sentry is yet
> another
> > >> clean pull of the release-1.3.0-rc2 tag (sha
> > >> : 31c8aca46c060685bd5a01f7706e2adab78a20d8); and
> > >> apache-sentry-1.3.0-incubating-src is the uncompressed version of the
> > tar
> > >> built in the previous step.
> > >>
> > >> However, when I try to generate a tar using  git archive --format=tar
> > >> --prefix=apache-sentry-1.3.0-incubating-src/ HEAD | gzip >
> > >> apache-sentry-1.3.0-incubating.tar.gz the resulting sha does not match
> > the
> > >> sha of the artifact generated by maven. When I unzip the resulting tar
> > and
> > >> diff it with the tar generated by maven the diff does come back clean.
> > >>  Does anyone have any thoughts on what I might be doing wrong here?
> > >>
> > >>
> > > I verified that the sha1 and md5 are different for tar.gz produced by
> > > maven versus git archive, although the contents are the same if I untar
> > the
> > > artifacts(diff -rq). My guess is they compress differently? @Greg, is
> > there
> > > a reason you used git archive rather than maven package?
> > >
> > > Ah never mind, checksums will be different for two different tar balls,
> > even if they are made from the same source. So we shouldn't really be
> > comparing checksums of two different tar balls to make sure their
> contents
> > match. Although we need to make sure the SHA and MD5 specified in the
> > release candidate match with SHA/MD5 of the tar. Does that sound right,
> > @Patrick, @Greg?
> >
> >
> > Also, as matter of voting protocol do we need to call a new vote or can
> we
> > >> continue to leave the existing vote open?
> > >>
> > >> Thanks,
> > >> Karthik
> > >>
> > >>
> > >>
> > >> On Fri, Apr 25, 2014 at 10:24 AM, Joe Brockmeier <[email protected]>
> > wrote:
> > >>
> > >> > -----BEGIN PGP SIGNED MESSAGE-----
> > >> > Hash: SHA1
> > >> >
> > >> > On 04/21/2014 11:33 PM, Patrick Hunt wrote:
> > >> > > I think Greg is right. It mostly looks good however there are a
> > >> > > number of significant differences between what's in the rc and
> > >> > > what's in the tag. Here's the diff output btw the untar'd rc and
> > >> > > what's in the tag in git (mostly iml files, but also some
> > >> > > directories, e.g. sentry-provider-db).
> > >> > >
> > >> > > I'm -1 at the moment.
> > >> >
> > >> > I've been hoping for some feedback from the release manager on these
> > >> > questions before proceeding to review the artifacts. Any progress,
> > >> > Karthik?
> > >> >
> > >> > Best,
> > >> >
> > >> > jzb
> > >> > - --
> > >> > Joe Brockmeier
> > >> > [email protected]
> > >> > Twitter: @jzb
> > >> > http://www.dissociatedpress.net/
> > >> > -----BEGIN PGP SIGNATURE-----
> > >> > Version: GnuPG v1
> > >> > Comment: Using GnuPG with Thunderbird - http://www.enigmail.net/
> > >> >
> > >> > iQEcBAEBAgAGBQJTWpppAAoJEKbW5zOgIHzUHfEH/3sSme6MLfTkU76fNJ4YKnFH
> > >> > Tlz6GCF+BrXruM+NRn1YtmaypM/briZK+pa9OIiH5kVNJ4w59VXgxQP8RPUnyAcm
> > >> > u7ZETekt6ioM8hUX71s/b7GPXceAhA6ZW3nzZSMbntdONsacjRMkwiCJ3Fz5buL8
> > >> > fmfx8ew0Zt7qrOYus0liwHZuE6CoTCu/a1nTTFZBPGpUr8ArsirO6mvcffK4YMX6
> > >> > /p6zAZgoOB1cc9bzQcdaMT79Hg7671HVsArY2I8XeG+g6kSPVYCsonDL/Kw1VHZw
> > >> > i41BulVIj/NU0HmPGRn65HIwdQHqbrakGjbVl9z9lCkyQ1QDtcVvUIj5B6iuabk=
> > >> > =nA9Z
> > >> > -----END PGP SIGNATURE-----
> > >> >
> > >>
> > >>
> > >>
> > >> --
> > >> Karthik Ramachandran
> > >> Mobile: 412-606-8981
> > >>
> > >
> > >
> > >
> > > --
> > > Sravya Tirukkovalur
> > >
> >
> >
> >
> > --
> > Sravya Tirukkovalur
> >
>



-- 
Sravya Tirukkovalur

Reply via email to