commons-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Ralph Goers <>
Subject Re: [parent] adding buildnumber in the manifest entries
Date Sat, 10 Sep 2011 16:21:56 GMT

Are you guys arguing about manifests or build processes. I can't really tell.

There is no perfect build process. I'm not a fan of voting on an RC and then renaming the
tag and so with VFS I created the tag over and over again for each candidate. That has its
own pitfalls but works since Nexus supports it well.  If we were following the Tomcat and
Httpd model I think the first release of VFS 2.0 would have been 2.0.20.  I would think that
would make it hard for users to understand what the latest version is, but if it works for
them I'd be interested to know more. Maybe their releases never fail.


On Sep 10, 2011, at 9:04 AM, sebb wrote:

> On 10 September 2011 16:28, Phil Steitz <> wrote:
>> On 9/10/11 8:18 AM, sebb wrote:
>>> On 10 September 2011 16:09, Phil Steitz <> wrote:
>>>> Could be I am misunderstanding the proposal here, but IIUC there is another
problem when it comes to release jars.  Current practice is to create the jars from what ends
up being the final RC tag.  This tag is then either copied or moved to the release tag, which
becomes the definitive source.  So at the time the jar is created, the tag/revision pair that
the build number should point to does not exist.  The only choice would seem then to be to
reference the RC tag which may end up deleted.  So again, it would seem better to me to either
omit the attribute from release jars or just use the (anticipated) final release tag name.
>>> Tag rename/copy history unfortunately shows the repo revision at the
>>> time of the copy which makes it quite hard to know what revision was
>>> actually used for the build. You have to scan the history to check for
>>> changes.
>>> The point is just to record what was used to create the jars.
>>> Yes, this will differ from the final tag, but having the information
>>> is better than not having it, IMO.
>> Not sure I agree.  Seems to me the whole point of having a release
>> tag - the final one that should really be immutable, is that it
>> points unambiguously to the source used to build the distribution.
> Agreed.
>> Why do we need to complicate things by referencing revision numbers
>> of (possibly deleted) RC tags?
> Because the official release tag was not used to build the project.
> The only way to know what was actually used is to record the tag and
> revision number.
> We don't have to store it in the jar manifests, but that seems as good
> a place as any.
>>> Note:
>>> One solution to all this uncertainty is to abandon RCs, and just use
>>> revision numbers.
>>> If the vote fails, bump the revision number and throw away the old
>>> revision and tag. This is what Tomcat and Httpd do.
>> Can you explain exactly how that works?
> This is the process AIUI:
> Tag 7.0.22 from trunk; build, vote.
> Vote succeeds - release 7.0.22.
> Vote fails, create tag 7.0.23 from updated trunk, build and redo vote
> Tag 7.0.22 could now be deleted, but I think they keep all the tags.
> Their achives [1] for 7.0.x don't contain 13, 17, 18; but the tags [2] do.
> [1]
> [2]
>> Phil
>>>> Phil
>>>> ---------------------------------------------------------------------
>>>> To unsubscribe, e-mail:
>>>> For additional commands, e-mail:
>>> ---------------------------------------------------------------------
>>> To unsubscribe, e-mail:
>>> For additional commands, e-mail:
>> ---------------------------------------------------------------------
>> To unsubscribe, e-mail:
>> For additional commands, e-mail:
> ---------------------------------------------------------------------
> To unsubscribe, e-mail:
> For additional commands, e-mail:

To unsubscribe, e-mail:
For additional commands, e-mail:

View raw message