A very excellent discussion of the issues by Scott has prompted me to
document my thoughts.
Ron
On 07/11/2014 5:24 PM, Scott Gray wrote:
I still think you're putting the cart before the horse. I'm not sure I could
name any specialpurpose component that has the two or three active contributors
that I think would be necessary to sustain it.
It will become apparent when the team or person steps forward to
maintain it.
If you have only one or two and they go away then what happens to any new
comers looking to contribute?
They will have to ask the leader of the sub-project. If the sub-project
community is completely unresponsive, they could propose to take over it.
Do we just make anyone who comes along a committer?
It is up to the sub-project to make that decision. Just the way Apache
leaves it up to OFBiz to decide who can commit.
The goal is to get these projects that have little or no current
support, set up in a way that they can find their own specialized
community that does not care to work on the framework.
Does the TLP PMC need to monitor a whole bunch of new mailing lists and then
support anyone who comes along in the hope they might be able to become
committers?
Not unless you care to stay on top of details. This is one of the
benefits of sub-projects.
You can reduce the flow through each list and only subscribe to stuff
you really need to follow, so you are only getting stuff that you
actually care about.
The archives are always available if you want to look up things after
the fact.
The sensible way to handle project management issues, is to set up an
OFBiz project mailing list that handles announcement or project news
that does not belong in the regular dev or user lists.
You can refer to the Apache project management pages for specific
suggestions about how Apache projects can handle this kind of activity.
If a component had no active committers how long would it be before it broke
completely because no one kept it updated?
Same as now. Soon.
Death of a module from natural causes ( no one caring) is not a bad thing.
It just would be much easier to know when a module had died.
Currently, the OFBiz community has a hard time knowing what has died and
what is still valued.
If a security vulnerability is found then who will deal with it?
Same people who deal with it now if they actually care to keep working
on the sub-project.
New people who actually use the module would probably be very interested
in getting high priority issues fixed.
At least, you would know who "should" be fixing it.
Much better than the current state.
It would be up to the sub-project to establish its policies about
support for security (back-porting, supported versions), etc.
Without active committers the PMC would need to remain familiar with the code
in order to receive contributions and determine their suitability for
inclusion. Nothing would be different from now except that the component would
be less visible to us, more easily forgotten and we'd have more infrastructure
to deal with.
Why would the framework committers even care?
It is up to the sub-project to sort out their own code integrity issues.
They have the detailed knowledge and the subject matter expertise and
the motivation (they actually use it).
If the project dies, you don't have to do anything except post a note on
the sub-project web site announcing that it is no longer active and that
the PMC is looking for a new team.
If anyone else from the PMC thought that this sub-project was important,
they would have joined the sub-project in the first place.
At least everyone would know that the module was no longer active.
You can look only at the best case scenarios because you won't have any
responsibility to these sub-projects. The PMC however will still have the
burden to do anything that the sub-project committers decide that they
themselves no longer want to do.
Why? If the module is useless and no one cares, why would the PMC do
anything besides documenting the situation on the sub-project web page.
"This project is no longer maintained. The code is available 'as is'.
The last release of the module is know to work with version 13.07. OFBiz
would be pleased to hear of a team willing to continue working on this
module."
Stop thinking about it after that until an individual or a team shows up.
At what point will it be okay by you for us to kill it off? What is it that
currently gives you confidence that it won't happen very quickly?
Yes, you can kill a project that no one is willing to work on or support.
At the moment it appears that several modules would have a hard time
surviving and should be killed.
From the discussion in this thread it appears that some are just tests
that failed.
This is better than the current case, wherein users have no idea about
what is active and what is dead and the PMC has to do a survey of all
users every time a decision about a dormant project has to be made.
I think Jacopo's solution is the best to keep everyone happy (including the
vocal minority) but IMO the best approach would be to remove the components
from OFBiz altogether, set them up on github or similar, ensure those who want
to take ownership have the required access and then advertise the existence of
these special purpose components from the OFBiz website. The components are
then free to stand or fall as they may. Should any actually survive and
thrive, then a subproject would seem like a good idea. It's my opinion though
that the vast majority of the components would go dormant and I don't think it
makes a difference whether it's as a subproject or as an external project
hosted elsewhere.
Removing them from Apache is a rather extreme approach that is likely to
lead to additional forking of the OFBiz project.
Apache has a well documented set of tools and processes for dealing with
modules or applications like this and there is no real value in
inventing some new process that is unlikely to be properly set up given
the amount of work being done in the supported bits of OFBiz.
I don't see any point of worrying at all about modules that have no
following.
We just need to be transparent about what is happening.
If they are supported, set up a sub-project that allows the interested
parties to work on them with as large a community as they can attract.
They would not have to get mixed in with the framework project that has
a different set of concerns and requirements for committing.
They might not be committers to the framework or core ERP
If the projects are not supported, leave them as is and state clearly on
the project web page that these modules are not supported.
If there is a history of people working on them, using them in
production and caring, it would be nice to document that, in case that
someone finds them useful.
If they were tests that failed before being finished, perhaps moving
them to a section of "dead test" modules in the SCM would satisfy the
desire not to lose any code regardless of its uselessness as well as
making it clear that these modules are not functional and users should
think that the name of the module has anything to do with what it will
actually do.
Mixing them in with modules that are actively maintained is not helpful.
I hope that this helps.
Ron
Regards
Scott
On 8/11/2014, at 9:21 am, Ron Wheeler <rwhee...@artifact-software.com> wrote:
Taher Alkhateeb raised some valid concerns.
My take (also as a newcomer) is that there is a high barrier to entry for
people working on the framework and this makes it hard to get new people into
the OFBiz project.
By creating sub-projects that have a much smaller scope and do not have any
affect on the overall robustness of the system, we would allow new people to
take on tasks that have a much narrower scope and are more in-line withtheir
abilities and interests.
It will also allow OFBiz to attract subject matter experts on certain areas
such as the BIRT language, data analysis, project management or manufacturing
who are not attracted to the framework development tasks.
The current level of complexity forces the group to be a small band of
dedicated people who have the detailed technical understanding of the framework
and tools used to build and maintain it.
There is nothing to prevent framework contributors from also joining
sub-projects where they have an interest.
It would also provide a level of transparency about what is getting supported.
If no one is active in the BIRT sub-project, at least you know that it is not
supported.
At the moment, you have no idea about the interest in supporting BIRT.
If you need it and it is not supported, currently you do not have anyone to
talk to except the framework mailing list.
If it had its own sub-project, you would have a leader and a list of people who
once had an interest in it.
If no one was interested in your BIRT issue, you could always hire someone to
work on the bits that you needed fixed.
If BIRT is completely orphaned,you could take over leadership of the BIRT
sub-project and get it back going.
I think that the project management and SCRUM projects could probably put
together sub-projects.
You would have to do a bit of work to get a BIRT group growing.
However, it looks like a good candidate for a separate project since BIRT is a
completely different programming language and a lot of the skills have to do
with business analysis, usability and data analysis rather than Java coding.
You might find that a BIRT sub-project attracts a number of people who would
not be interested in supporting the framework.
Sub-project will also reduce the amount of traffic on the dev list and allow
people to focus on what they think matters to them.
They also allow other people to take on leadership roles in these areas which
reduces the burden on the current core contributors.
Sub-projects are a key part of many Apache projects, so I believe that they
must serve a useful purpose.
I think that this project is really in need of a way to grow the community
without diluting the quality and I see sub-projects as a way to keep the focus
within Apache OFBiz rather than fork the parts into outside open source
projects which is the current direction.
Ron
On 07/11/2014 2:08 PM, Taher Alkhateeb wrote:
Hi Everyone,
I do not have a long history with the OFBiz project to understand its history,
but from the last few years I noticed the following:
- The system is huge
- Documentation is wanting
- The community is suffering from low number of experienced developers
For example, I use and want to support the BIRT reporting component. Yet there
are very few committed developers experienced and comfortable enough in
maintaining it and upgrading with new releases. So, I would imagine taking it
out as an almost sure death sentence given the already low resources.
With all due respect, talking about sub-projects and plans and resources and commit
access and all of that stuff does not make sense when you barely have enough experienced
people maintaining the code. I see only a few names over and over who are doing the
"real" heavy work.
So for my 2 cents, I still strongly encourage the initial suggestion by Jacopo.
I think other suggestions would probably kill any less heavily maintained
components.
Taher Alkhateeb
----- Original Message -----
From: "Ron Wheeler" <rwhee...@artifact-software.com>
To: dev@ofbiz.apache.org
Sent: Friday, 7 November, 2014 9:29:05 PM
Subject: Re: How to use ProjectMgr in 13.07
I was trying to find some Apache docs about what is involved.
Separating the SCM controls so that the sub-projects can have their own
committers is an important task.
Any idea about what else is required?
In any case, it would be the people who want to support the sub-project
to do the paperwork.
There is clearly nothing to stop a fork of any part of OFBiz but there
is some advantage to keeping OZBiz in one piece.
The most important thing is that it allows the sub-project to use the
OFBiz name and Apache branding in its "marketing" material and
documentation.
It also builds the pool of potential contributors and committers for the
core.
Ron
On 07/11/2014 11:44 AM, Jacopo Cappellato wrote:
I am fine with the idea of encouraging the growth of an ecosystem of *projects*
about OFBiz (not necessarily all within the ASF) but I disagree that they
should be *sub-projects* of OFBiz, mostly because sub-projects just add
complexity within the OFBiz community (with more paperwork required).
Jacopo
On Nov 7, 2014, at 5:32 PM, Adrian Crum <adrian.c...@sandglass-software.com>
wrote:
I agree with a separate community approach, for these reasons:
The special purpose components started out as little demonstrations of how
OFBiz can be extended to role-specific applications. Since then, some of those
components have expanded into full-featured applications - so the overhead of
maintaining them has increased beyond what we expected initially.
Some special purpose components were included as the result of a community discussion and
effort, but others were simply "dumped" into the repository without any
discussion or community participation - and as a result they receive little support.
Some special purpose components were created and initially supported by
community members who are not around any more.
As a result of all of these things, the special purpose components are not well
maintained.
From my perspective, if anyone believes a component needs more attention and
would like to develop it further, then they should spin it off to a separate
project.
So, instead of having a conversation about how the OFBiz community can better
support the Project Manager component, maybe we should have a conversation
about how to spin it off to a separate community.
Opentaps started off that way. Initially, it was OFBiz plus additional
components: Financials, CRM, and Warehousing.
I think the OFBiz community would benefit if we stopped trying to be all things
to all people, and instead focus on providing a sound and flexible data model -
combined with robust, reliable services. Special purpose applications, and even
presentation layer details can be left to other communities.
Adrian Crum
Sandglass Software
www.sandglass-software.com
On 11/7/2014 4:02 PM, Ron Wheeler wrote:
I may be beating a dead horse but what Jacopo is proposing and the
concern that Jacques raised about resources would seem to fit very well
into a sub-project structure.
Split these modules out of the main line into their own OFBiz
sub-projects where they could attract their own resources (committers
even) and would be responsible for delivering modules that worked with
the various OFBiz cores that exist.
Let the sub-projects worry about their own relationship to OFBiz
releases and release branches.
They might just support the released 13.07.xx version, if that is what
the sub-project's user community can support or they might maintain 2
versions 13.07 and a 14.xx. that works with a recent version of the trunk.
In any case, it would no longer be a problem for the OFBiz core
developers and they could release just the OFBiz components that are
officially part of the core.
Clearly good communication between the core project and the sub-project
about release plans would help everyone but the core group would be
basically free to release the core as they want and the sub-projects
would have to communicate to their uses communities what impact this
would have on the users of the modules.
If a sub-project needs a change to the core to implement some feature,
they would have to file an enhancement JIRA issue and convince someone
to add it (unless they are a committer on the core).
If two sub-projects had a compatibility issue, they would at least know
who to talk to get a solution.
Ron
On 07/11/2014 7:04 AM, Jacopo Cappellato wrote:
On Nov 7, 2014, at 12:36 PM, Jacques Le Roux
<jacques.le.r...@les7arts.com> wrote:
This will no longer work for some components (scrum for instance)
I believe we could be reintroduce some specialpurpose component in
next release,
There is a difference between including them in a release branch and
including them in the releases: I would be more inclined to include
(all of them) in the release branches but exclude them from the
releases; this would simplify the work required to keep them in synch
and would also help end users to integrate them.
However the specialpurpose components should be disabled in order to
avoid the users to get a default ootb release/branch with enabled
special purpose functionalities that may override the more general
purpose ones offered by the core applications.
We should still consider the idea of providing separate products for
the specialpurpose components (and having them in the branch would
help this process).
If the idea I am proposing here (include the specialpurpose components
in the branch but not in the releases) we could re-add them (as
disabled) also to the 13.07 branch but exclude them from all the
releases (13.07.02 etc...): this will protect all the stabilization
work we did on the branch (and also from some licensing issues that
may affects some of the artifacts in some of the specialpurpose
components) .
Jacopo
as long as they are backed by some efforts, come to mind
project manager (Pierre Smits?)
scrum (Hans?)
examples and ext (at least me)
myportal (French people use portals, not sure for myportal?)
Other components?
IRRW Jacopo said he was not against a new discussion on this subject
(I could not find his message), what do you think?
Jacques
Le 21/10/2014 09:06, gil portenseigne a écrit :
I've never used svn external property, just discovering. That sounds
usefull and i'll try it out !
Thanks for the advice !
Gil
On 20/10/2014 19:08, Jacques Le Roux wrote:
I use svn external in the stable demo, already explained that in
the MLs: see
https://svn.apache.org/viewvc/ofbiz/trunk/tools/demo-backup/branch13.7-demo.patch?view=markup
You can use the same to keep in sync, only consider projectmgr in
your case. Since there is no projectmgr in R13.07 the risk of
gettins conflicts or build issue is extremely low
Jacques
Le 20/10/2014 17:28, gil portenseigne a écrit :
Hi Jacopo,
Ok then, i will have to re-synchronize new trunk devs each time
i'll feel it necessary. My fear is about incompatibility between
13.07 and trunk technologies, but that won't happen soon, or i
might backport the evolution into my local environment.
That will do the job !
Thanks
Gil
On 20/10/2014 16:36, Jacopo Cappellato wrote:
Hi Gil,
I would suggest to check it out from the trunk to the hot-deploy
folder of 13.07:
cd hot-deploy
svn co
http://svn.apache.org/repos/asf/ofbiz/trunk/specialpurpose/projectmgr
Jacopo
On Oct 20, 2014, at 4:11 PM, gil portenseigne
<gil.portensei...@nereide.fr> wrote:
Hi all,
I don't want to relaunch the debate around including the
projectMgmt component into the 13.07 release, but i have a
question :
What is the best way to import the projectMgr component in my
local 13.07 checkout environment, to start using it in a real
project and to contribute on upgrading it for trunk and/or the
13.07 release ?
Trunk technical evolution might be a problem if a want to stick
to 13.07 release with projectMgmt features.
Should I use trunk instead ?
Cheers
Gil
--
<siteon0.jpg>
Gil Portenseigne
Consultant ERP OFBiz
Société Néréide
3b Les isles
37270 Veretz
Tel : 09 74 53 46 09, puis 1, poste 61
Mob : 06 82 740 444
www.nereide.fr
--
<Mail Attachment.jpeg>
Gil Portenseigne
Consultant ERP OFBiz
Société Néréide
3b Les isles
37270 Veretz
Tel : 09 74 53 46 09, puis 1, poste 61
Mob : 06 82 740 444
www.nereide.fr
--
Ron Wheeler
President
Artifact Software Inc
email: rwhee...@artifact-software.com
skype: ronaldmwheeler
phone: 866-970-2435, ext 102
--
Ron Wheeler
President
Artifact Software Inc
email: rwhee...@artifact-software.com
skype: ronaldmwheeler
phone: 866-970-2435, ext 102