Home » Language IDEs » ServerTools (WTP) » Call for potential contributors / survey (was: Enough people to get started with)
| Call for potential contributors / survey (was: Enough people to get started with) [message #13619] |
Thu, 15 January 2004 03:18  |
Eclipse User |
|
|
|
If you a potential contributor please summarize your opinions and vote
for existing project or new project. If you vote for new project -
please propose names too... And who do you think will _own_ that new
project (not contributed code) - it is important because if noone take
the responsibility project would newer even created!
Make a note if you already have something to contribute, if not - what
do you like to do for a project...
And as a start it is my vote...
----------------------------
Igor Malinin
* project: SolarEclipse (http://solareclipse.sourceforge.net/)
* if new project decided i prefer abstract name (with no exact proposal,
let others think about name this round)
* coontribution: Anything from SolarEclipse under CPL
|
|
|
| Re: Call for potential contributors / survey (was: Enough people to get started with) [message #13744 is a reply to message #13619] |
Thu, 15 January 2004 10:35  |
| Eclipse User |
|
|
|
Igor Malinin wrote:
> If you a potential contributor please summarize your opinions and vote
> for existing project or new project. If you vote for new project -
> please propose names too... And who do you think will _own_ that new
> project (not contributed code) - it is important because if noone take
> the responsibility project would newer even created!
I'd like to add my thoughts here. You are pointing out a very important
-- but nevertheless often neglected aspect -- of open-source
development: Who gets to make the decisions? It's improbable that all
contributors to a project will always agree on everything... be it code,
infrastructure, politics or even people. Often issues can be discussed
until a concensus is reached, but sometimes that's not enough. In any
project that has more than one regular contributor, basic rules for
decision-making need to be established and agreed upon.
Igor, you are suggesting a model where one person "owns" the project. On
the thread "Enough people to get started with" you said:
"There must be some owner. Or the project could be destroyed
easily. But project owner is different to code owner. Project
owner is a person or group who allow or disallow other people
to contribute. Any project, commercial or opensource, has an
owner."
I understand where you're coming from. I've been involved with several
open-source projects in the past couple of years, some of which have
been successful, while others have been utter failures. I've also been
closely watching many open-source projects that I wasn't directly
involved with, and tried to figure out why they were succeeding -- or
failing.
The one model that I think works extremely well is the model of
"meritocracy" used, most prominently, by the Apache Software Foundation
<http://www.apache.org/>. At Apache, those who are most active are
rewarded by the community with responsibility. You don't get into a
project just by saying "I've got 13 years of experience in developing
large software systems, and I'd like to work on YYY for this project,
can you give me write access to the repository". Not. Instead you start
contributing to the project by joining discussions, helping out users on
the mailing lists or forums, submitting patches to fix bugs or even add
small features. Sooner or later the existing project members will know
your name and respect your opinions. If you continue to contribute in
this way over a reasonably long time span (like one or two months),
someone will call for a vote to give you the status of a "committer",
whereby you get write access to the CVS repository and the right to
participate in votes (this is not 100% correct, but you get the idea).
So if enough members agree that you're a cool person to work with,
you're in.
The members get to vote on other things, too. For example, any committer
can veto a change after reviewing the CVS commit log, maybe because they
think you haven't done this in the right way, or whatever. You back out
your change and try to resolve the issue with the other team members on
the mailing list. Maybe you can convince the vetoer that you were right,
or you'll find a compromise together. Project members also vote on
releases, roadmaps, etc.
The important thing is that the more you do (and do well), the more
respect and faith you will earn from the team, and the more
responsibility and influence on the future of the project you are
granted. This model serves one important aspect of volunteer-backed
development very well: people come and go. Often the set of active
members continuously changes, and the meritocracy model takes that into
account.
Now, the hard thing is how to bootstrap such a project when there are
multiple initial code contributions. Few or even none of us have worked
together yet, so we haven't had much of a chance to build "trust
relationships". But if we can actually come to a consensus concerning
basic things like the name, the design goals, the infrastructure and a
development model, we have a good chance on being able to collaborate
after the project has been launched.
And in case anyone was wondering: yes, there is room for people who
contribute other resources than code: if someone works on the
infrastructure, writes documentation, manages releases and
announcements, etc etc, that can also be rewarded by team membership.
> Make a note if you already have something to contribute, if not - what
> do you like to do for a project...
So here we go:
As should be obvious now, I'd be happy to donate the CSS code from
<http://csseditor.sourceforge.net> (assuming we can come to an
agreement, of course).
My vote would be for a new project with a functional name, something
like "Web Development Tools for Eclipse". My favorite qualified name and
URL for the project would have been 'wdt.sf.net', but I just realized
that that has already been registered by some freaky FreeBSD driver
project. So what about 'wdte.sf.net', 'ewdt.sf.net', or maybe
'eclipsewdt.sf.net' (keeping in mind that those would also be used as
package names, unless someone wants to donate a real domain).
I vote for a functional name because (a) I don't have a good idea for an
abstract name, (b) I believe a short, functional name that is easy to
memorize may help adoption of the basic plugins by other projects and
(c) it makes it the logical candidate for promotion into the Eclipse Web
Tools project, if that ever becomes real. I also think that an abstract
name doesn't fit with a project that provides basic features for an
existing platform. This is not going to be a research project
experimenting with intentional programming or any such "nonsense" ;-)
And I don't buy the argument that an abstract name would give the
project more "freedom to innovate" (to quote M$): IMHO the project
whould have a clear focus on web technologies, which is already broad
enough.
For project management, I explained my preference for the so-called
"meritocracy". Non-acceptance of that model would probably be a
showstopper for me. I also would volunteer to take over some
infrastructure tasks like getting the project and website setup. I've
just been through the whole process for csseditor.sf.net, and the memory
of it is still pretty fresh ;-)
Enough for now, this post is already far too long. Thanks if you've come
this far :-)
Cheers,
Chris
--
cmlenz at gmx.de
|
|
| | |
| Re: Call for potential contributors / survey [message #13966 is a reply to message #13784] |
Thu, 15 January 2004 18:13 |
| Eclipse User |
|
|
|
Sjoerd van Leent wrote:
> The model of responsibility you suggested does have a point. But at the
> beginning, no-one has contributed code, design or theory about the
> project. So you need to have someone who cuts through the ideas and makes
> decisions.
The whole point of this discussion is that there *are* people willing to
contribute code that was developed independently. Those people should
also get to be the ones that decide what should happen with the code.
Frankly, because they have done all the work. There isn't going to be a
big upfront design marathon. We're just going to continue to work on the
code, only that there will be more eyes on the individual parts, and
there will be cross-pollination.
> Also I think if the project grows larger at the time, we need to have
> financial backup (most likely from a mix of two or three settled companies
> and a bunch of smaller companies).
One thing at a time. We're light years away from being at that point.
Really, what do we have? Editors for XML/HTML, CSS and JavaScript with
only basic features. And even that is only if the contributors (which is
pretty much just Alex, Igor and myself) can agree to collaborate.
Everything else is hot air.
If the project ever gets as large and successful as you seem to expect
(and that's a huge "if"), I'd personally be disappointed if it wouldn't
have been invited to join the Eclipse foundation.
> We should have a core group of software engineers that lead developers to
> making there piece of the project work. But we also need a core group of
> information engineers who can bring the product to clients and can
> convince companies to give support (be it financial support or product
> support).
>
> I guess what I am trying to say can be represented in the following
> structure:
[snip]
> I think that everyone should have experience with development and bug
> hunting before going to Software Engineering. If the project really
> explodes it is not unusual to have even more levels and a complete staff.
> Note that information engineers do not necessarily have people below them,
> but are at the same level as software engineers.
Whoa, we're totally out of sync here. You are describing an organization
that would work for companies, but I don't think it'd ever work in
open-source development. You can't force anyone to do the work you think
is needed, there's no delegation, you either do it yourself or wait
until one of the other developers gets to it. If you wonder how that
could possibly work, have you by any chance heard of Apache-HTTPD,
Tomcat or Struts?
I'd be curious to hear what your background in open-source development
is, like what kind of projects you have been involved with.
Really, if there's gonna be something like a president or chief software
engineers that delegate the dirty work to meer developers, I'm out.
Cheers,
Chris
--
cmlenz at gmx.de
|
|
| | | | |
| Re: Call for potential contributors / survey [message #13989 is a reply to message #13978] |
Sat, 17 January 2004 07:02 |
| Eclipse User |
|
|
|
Sjoerd van Leent wrote:
> Some answers though not complete...
>
> Currently I am working on an implementation of a Linux based component
> oriented distribution.
>
> Although the organogram seems very comprehensive it is not that difficult.
> All I want to make clear that without a decent lead engineer (owner)
> people will fall into a "my-self-made-code mode" where it must be a
> "we-create-the-code mode." I am only suggesting that. I am not saying that
> the owner shouldn't be chosen out of the contributors, actually the
> reverse, I think this should be done. However, someone should do the
> kick-off, again, this can be a elected engineer/developer.
>
> I do think that we need to see the larger picture at this point. What are
> we going to make?
> - Are it just some components which actually do nothing with each other?
> - Is it a framework for adding components to (which could be implemented
> by ourselves and/or by third parties)?
> - Or is it a total product?
Again, I'm just gonna state my personal opinion, and am not talking for
anyone else:
I'd be in favor of a project that created plugins for the Eclipse
platform that would together provide an integrated development
environment for standard web technologies. Currently I don't see very
much except editors and validators fall into that category, although
features like -- for example -- HTTP protocol tracing, WSDL inspectors
and stuff like that would also fit in.
And yes, the different plugins would work together. So for example, the
XHTML plugin would have a dependancy on both the CSS plugin as well as
the JavaScript plugin. Whenever the XHTML editor encountered a <style>
tag with contents of type text/css, it'd delegate handling of that range
of the document to the CSS plugin. Whenever the XHTLM editors
encountered a <script> tag with type text/javascript, it'd delegate to
the JavaScript plugin. And so on. If those plugins would not be
installed, it'd just handle the contents of those tags as plain text.
These plugins would be bundled into an installable Eclipse feature. But
the plugins would not only be able to work with each other, it should
also be possible to use them in third-party plugins. So if you have a
JSP editor, everything just moves one level down. The JSP editor, if the
content type of the page is text/html or text/xml+html, would delegate
to the HTML or XHTML plugin to handle the template text of the page.
You can of course consider this a complete product. Download the Eclipse
platform and install the web development tools feature, and you'll have
editors for standard web technologies, all tightly integrated and with
cool features such as versioning provided by the platform. But the
product can still be built upon, and for many people, it'll only be
another part of their installation.
Cheers,
Christopher
--
cmlenz at gmx.de
|
|
|
| Re: Call for potential contributors / survey [message #13991 is a reply to message #13989] |
Sat, 17 January 2004 08:45 |
| Eclipse User |
|
|
|
Christopher Lenz wrote:
> I'd be in favor of a project that created plugins for the Eclipse
> platform that would together provide an integrated development
> environment for standard web technologies. Currently I don't see very
> much except editors and validators fall into that category, although
> features like -- for example -- HTTP protocol tracing, WSDL inspectors
> and stuff like that would also fit in.
> And yes, the different plugins would work together. So for example, the
> XHTML plugin would have a dependancy on both the CSS plugin as well as
> the JavaScript plugin. Whenever the XHTML editor encountered a
<style></style>
> tag with contents of type text/css, it'd delegate handling of that range
> of the document to the CSS plugin. Whenever the XHTLM editors
> encountered a <script></script> tag with type text/javascript, it'd delegate
to
> the JavaScript plugin. And so on. If those plugins would not be
> installed, it'd just handle the contents of those tags as plain text.
> These plugins would be bundled into an installable Eclipse feature. But
> the plugins would not only be able to work with each other, it should
> also be possible to use them in third-party plugins. So if you have a
> JSP editor, everything just moves one level down. The JSP editor, if the
> content type of the page is text/html or text/xml+html, would delegate
> to the HTML or XHTML plugin to handle the template text of the page.
> You can of course consider this a complete product. Download the Eclipse
> platform and install the web development tools feature, and you'll have
> editors for standard web technologies, all tightly integrated and with
> cool features such as versioning provided by the platform. But the
> product can still be built upon, and for many people, it'll only be
> another part of their installation.
I don't think that we really have so much different ideas about it, we
only have some different points where the system should be complete and
where it should be less complete. Which in my idea could be a good thing.
We do agree that the system should be flexible though comprehensive (e.g.
not just some loose parts) and it should give web-developers a base
environment (discussions about "what is a..." can be somewhere in the near
future). The system also should provide services to implementors of other
products based on this environment which are usable, understandable and
flexible (such as the delegation you did spend some words about).
The only point where we are having some differences is that I suggested a
product which is basically an framework which provides these services
where upon a basic standard could be build and that you suggested a
framework with basic support and forms a product on it's own, though
flexible enough to be used by other implementors (preventing hunderds of
XML editors for example and other kinds of useless reinventions of the
wheel).
The idea of having a basic product build upon the framework honors the
framework, at least, that's my opinion. So we want pratically both the
same sort of system, however, we just misinterpret eachother.
|
|
|
| Re: Call for potential contributors / survey (was: Enough people to get started with) [message #14001 is a reply to message #13982] |
Mon, 19 January 2004 11:10 |
| Eclipse User |
|
|
|
Originally posted by: adam.fowler.ideagen.co.uk
"Mathew Brozowski" <mbrozowski@nc.rr.com> wrote in message
news:bu9ifp$nc2$1@eclipse.org...
> Not to be merely a dissenter... But as for names I think I'd prefer webdt.
> Seems to match the
> Eclipse Xdt style a little better and is clear for the web. wdte is
really
> hard to guess if you don't
> already know it.
I would agree with this! Sorry all. 8o)
Webdt does sound better, and is more consistent.
I am currently working for a company who uses NetBeans. I would personally
love an excuse to move to Eclipse. (NetBeans *really* annoys me.) In order
to do this, however, I would need some complex(?) components. Namely, some
form of Bean creation wizard(s), especially for EJB 2 Container Managed
Relationships (CMR). I hate editing XML, and would really like to see any
errors before deploying to a server!
In order to build these wizards, certain functionality would be required.
Please note that I haven't developed for Eclipse before so have no idea
about the current API.
- Ability to override the XML editor for certain files - namely ejb-jar.xml
and have this vanish from the file view and replaced with something like
'EJB module'. This would allow me to develop a component that tracked
changes to Bean classes and dynamically maintain a list of business methods,
schema mappings, etc. in order to update the ejb-jar file.
- This also needs a read-only XML editor view to check settings visually
for convenience.
- EJB jar file generation using Ant. Providing custom input from above ejb
component.
- Plugin to a common DB access API for setting up project wide DB
connections (For automatic schema creation) within the EJB wizard.
- Set of verification tools built on top of above functionality. This would
happen prior to or part of the Ant-based build and deploy process.
- This does imply a common server connect/deploy framework to base this
component upon
Can't think of anything else right now, basically if you have access to it,
look at the EJB wizards in the Sun ONE IDE. I have no idea if the normal
NetBeans IDE download contains the relevant code, but its the only redeeming
feature of NetBeans I've found.
Some sort of EJB design analysis component would also be good, but that
basically would build on top of the above functionality.
I'm quite hapy to code this in my spare time (I hate NetBeans THAT MUCH!).
I'm also quite happy to act as a general dogsbody regarding web site
maintainence, project correspondence admin, tracking which developers may be
'worthy' of contributor status, etc.,etc. I like the sound of this project,
if only we could get on with it!
To help with that, how about getting someone to act as an 'arbitrator'
initially. This person would basically answer to all the individual
contributed component developers and would act as a central point of
reference and management. They would act without veto though, so just
another 'contributor' but on the management side of things. Any thoughts.
Sorry the the stupidly long post, I'm a gobby so-and-so.
Regards,
Adam.
--
Adam Fowler BSc
http://www.adamfowler.org/
|
|
|
| Re: Call for potential contributors / survey (was: Enough people to get started with) [message #14348 is a reply to message #13984] |
Tue, 20 January 2004 02:09 |
| Eclipse User |
|
|
|
I see quite a ground covered by opinions of so many people, which is
heartening and I am hoping a development effort will emerge out of these
discussions.
I am all for it ,and here's my vote (formally !!) . Name doesn't matter to
me (frankly) as long as I have something to contribute to. I'd like to see
some tools for J2EE development(JSP, EJB, JMX, XML, Webservices...etc etc).
Cheers,
Rahul
|
|
|
Goto Forum:
Current Time: Wed May 07 18:21:25 EDT 2025
Powered by FUDForum. Page generated in 0.03727 seconds |
Home