Home » Language IDEs » ServerTools (WTP) » Milestone Plan
| | | | |
 Re: Milestone Plan [message #47031 is a reply to message #46913] |
Wed, 29 September 2004 23:21   |
Eclipse User |
|
|
|
I agree with Dave. Let's see how the project has been going so far...
This project has been talked about for well over a year. IBM had code
running and proposed a contribution in February, and we talked about it at
EclipseCon 2004. Actual contribution was delayed until we could talk about
the group organization then until an Eclipse director could be picked. We
talked a lot about charters and scopes and commoditization and flexible
projects and whatnot. Work was delayed until the project could be officially
approved.
On July 22nd both the IBM and Lomboz contributions were posted in a zip file
format, useful for trying it out but not for doing any development with it.
Interest was so high that the eclipse.org web site was brought down by
people trying to download the zips. Pushing to CVS was delayed until the
team could talk about what directory and package names to use. But when a
few small pieces were finally pushed to CVS, the package names were wrong
anyway. :) That's Ok, things go wrong; it's not a big deal as long as you
can iterate and fix them quickly. But now, according to the new milestone
plan, it's hoped that there will be some editors working in 16 weeks, and
the SQL viewer in 24 weeks.
Meanwhile the talented developers at Genuitec, not burdened by any red tape
or other bureaucratic issues, released a product using large parts of the
IBM contribution on August 16. This is a good thing; they are *supposed* to
productize the code, remember? But look, it only took them 25 days, counting
weekends, from when the zip was available. This is even more impressive when
you realize that to have a final, stable commercial GA product release on
that date they had to have something compiling and running well before then.
How long did that take, 5 days, 10 days?
So on the one hand I see a large cumbersome open source project plodding
along to do something in ~6 months what a small motivated group of agile
programmers did in ~6 days. What is wrong with this picture?
The developers who contributed all this good code must be incredibly
frustrated to see this drag on for so long. I know I would be. Clearly the
potential users are frustrated too. I apologize if this sounds overly
critical but this project has been, and continues to be, an embarassment to
Eclipse. It's so important, and there are so many talented people begging to
work on this and contribute to it. So what is the problem?
I don't care to speculate, but here is a simple plan to fix it. Deciding on
package names and directory structures shouldn't take more than, say, a
week, even if you're as long winded as I am. So one week from now, we would
like to see every file from the IBM contribution, including any changes made
in private IBM repositories since July, pushed to eclipse.org public
webtools CVS. Builds should start immediately, by hand if necessary until
the automated scripts can be set up. That's the first milestone. One week
later, any changes made by others (such as Genuitec) since the initial
contribution should be merged into public CVS. At this point there should be
no private repositories, only the one true source at eclipse.org. Future
milestones should be synced up to coincide with Eclipse Platform 3.1
milestones and follow the excellent project management techniques pioneered
by the Platform team. Code and concepts from the Lomboz contribution and
anywhere else can be added on an ongoing basis. Simple.
I challenge the current leadership of the webtools project to implement this
plan immediately. Don't waste time explaining in great detail why it's not
possible or whatever, just try it. Do I think it will happen? The cynic in
me says no, but the optimist says maybe we'll be surprised. (Usually the
cynic is right. :)
--
Ed Burnette
www.eclipsepowered.org
Only personal opinions need apply.
"Dave Carlson" <dcarlson@xmlmodeling.com> wrote in message
news:cjelua$jgq$1@eclipse.org...
The plan shows that the first structured editor for XML will not be
available until M2. If the milestones are 8 weeks apart (per the plan) and
assuming we are at M0 now, then it will be another 16 weeks before we see a
milestone with the first structured editor.
.. . .
|
|
| | |
| Re: Milestone Plan [message #47221 is a reply to message #46913] |
Thu, 30 September 2004 17:30 |
| Eclipse User |
|
|
|
Originally posted by: dominique.devito.objectweb.org
Dave Carlson wrote:
> The plan shows that the first structured editor for XML will not be
> available until M2. If the milestones are 8 weeks apart (per the plan) and
> assuming we are at M0 now, then it will be another 16 weeks before we see a
> milestone with the first structured editor.
> If this is the case, I will use the IBM contrib source for XML & XSD editor
> to fork a private development branch to add bug fixes and create a
> distribution and update site for my co-workers.
There are people and organizations working on the Structured Source Editor
(SSE) : Innoopract, OpenWide, IBM, Jens Lukowski...
The current task of these people is to review IBM code, and to put
selected code pieces into CVS. Then, this starting amount of code will be
available for refactoring, improvements and so on.
Instead of forking, why don't you join these people to work on SSE ? Any
contribution is very welcome ! Just tell us.
> Can we get an initial CVS structure populated with the XML structured editor
> code before M2 work starts?
Well, the XML structured editor code is expected for M2. This implies that
some code in CVS will be available before M2 and will be worked to achieve
M2 goals.
And as I have heard, some SSE parts are in preparation for moving into
CVS. Then, you will have an initial CVS structure populated for SSE.
Dominique
> Thanks,
> Dave Carlson
> "Dominique De Vito" <dominique.devito@objectweb.org> wrote in message
> news:cje9h1$lmc$1@eclipse.org...
> >
> > Done. The milestone plan is published - I am one day late due to WebDAV
> > access (own) pb :-(
> >
> > The milestone plan :
> >
http://www.eclipse.org/webtools/development/planning/milesto ne_plan_1_0.html
> >
|
|
|
| Re: Milestone Plan [message #47277 is a reply to message #47221] |
Thu, 30 September 2004 22:43 |
| Eclipse User |
|
|
|
Hi Dominique,
Thanks for your reply. I am interested in contributing to the XML and XSD
editors (in fact, if you were at the WebTools BOF at EclipseCon, I was the
one who raised a comment about wanting to work on refactoring support for
the XSD editor, i.e. refactoring XSD constructs).
I also want to look into contributing some of my prior work in XML modeling
(probably more in design than in code at this point, given the different
code base). See my web site at www.xmlmodeling.com.
I had no idea who (if anyone) was working on the sturctured editor up until
now. With nothing in CVS, it was not apparent that there was anything to
contribute to...
Thanks,
Dave Carlson
"Dominique De Vito" <dominique.devito@objectweb.org> wrote in message
news:cjhtuc$nln$1@eclipse.org...
> There are people and organizations working on the Structured Source Editor
> (SSE) : Innoopract, OpenWide, IBM, Jens Lukowski...
>
> The current task of these people is to review IBM code, and to put
> selected code pieces into CVS. Then, this starting amount of code will be
> available for refactoring, improvements and so on.
>
> Instead of forking, why don't you join these people to work on SSE ? Any
> contribution is very welcome ! Just tell us.
>
|
|
|
| Re: Milestone Plan [message #47315 is a reply to message #47277] |
Fri, 01 October 2004 03:58 |
| Eclipse User |
|
|
|
Originally posted by: dominique.devito.objectweb.org
Hi Dave,
Dave Carlson wrote:
> Hi Dominique,
> Thanks for your reply. I am interested in contributing to the XML and XSD
> editors (in fact, if you were at the WebTools BOF at EclipseCon, I was the
No, I was not there.
> one who raised a comment about wanting to work on refactoring support for
> the XSD editor, i.e. refactoring XSD constructs).
> I also want to look into contributing some of my prior work in XML modeling
> (probably more in design than in code at this point, given the different
> code base).
You are very welcome !
> See my web site at www.xmlmodeling.com.
Yeah, it looks very interesting ! I am also taking here my hat 'UML' as I
am going to do more and more UML with MDA approach (on other projects).
> I had no idea who (if anyone) was working on the sturctured editor up until
> now. With nothing in CVS, it was not apparent that there was anything to
> contribute to...
Due to important code contributions for Eclipse WTP, we have starting with
code reviews. Since some weeks (let's say to simplify: the beginning of
September), there are growing working groups on different topics.
Since the beginning of the week, I feel the need to publish on eclipse.org
information about working groups to let the community know who is working
on what. Plus entry points (emails for contact).
I expect, then, to publish information this evening or in the WE.
Dominique
> Thanks,
> Dave Carlson
> "Dominique De Vito" <dominique.devito@objectweb.org> wrote in message
> news:cjhtuc$nln$1@eclipse.org...
> > There are people and organizations working on the Structured Source Editor
> > (SSE) : Innoopract, OpenWide, IBM, Jens Lukowski...
> >
> > The current task of these people is to review IBM code, and to put
> > selected code pieces into CVS. Then, this starting amount of code will be
> > available for refactoring, improvements and so on.
> >
> > Instead of forking, why don't you join these people to work on SSE ? Any
> > contribution is very welcome ! Just tell us.
> >
|
|
|
| Re: Milestone Plan [message #47353 is a reply to message #47031] |
Fri, 01 October 2004 09:15 |
| Eclipse User |
|
|
|
Originally posted by: jkrause.w4toolkit.com
Ed,
It is pretty difficult to argue that we made a great job in the past :-(
but I don't feel comfortable with your proposed plan.
Here are my reasons:
The IBM contribution is huge, and as soon as we import the code into CVS
we have a commitment to maintain and improve that code. It is very hard
- next to impossible to take away something that was around before.
The API of the IBM contribution is huge, and it will be almost
impossible to reduce it in the future after people have started
extending it. This means that extending the web tools platform will
remain complex for all future, something that I would like to avoid.
Web Tools is intended to be an extensible platform plus examplary end
user tools, not as only end user tools. This requires more thoughts and
careful steps then just creating a status quo with a single code
contrtibution. Everybody is free to create a repository somewhere to
produce end user tools based on IBMs initial contribution. But don't get
me wrong: I don't want to shy away possible contributors, and I agree
that we need to have code in CVS very quickly.
So here is what I would prefer to do:
Lets extract the smallest working “nucleus” from the different parts of
the IBM contribution, and then try to extend every nucleus with new
functionality. This may be sample implementations or existing parts of
the IBM contribution. That way we can document the API / SPI that is
needed to use / extend core functionality.
With working code, documentation and maybe some sort of tutorials that
we should create while executing the work described above we will get a
lot of community feedback as well as real working software and probably
/ hopefully more contributors as it becomes easier for them to get started.
I have been working in the past weeks to set up a team to start this
process for the structured source editor and we are mainly awaiting CVS
committer rights to move the first results of our work into CVS. So
hopefully you will see core sse functionality in CVS.
Jochen Krause
Innoopract
The above is my personal point of view and does not necessarily
represent the point of view of the PMC.
Ed Burnette wrote:
> I agree with Dave. Let's see how the project has been going so far...
>
> This project has been talked about for well over a year. IBM had code
> running and proposed a contribution in February, and we talked about it at
> EclipseCon 2004. Actual contribution was delayed until we could talk about
> the group organization then until an Eclipse director could be picked. We
> talked a lot about charters and scopes and commoditization and flexible
> projects and whatnot. Work was delayed until the project could be officially
> approved.
>
> On July 22nd both the IBM and Lomboz contributions were posted in a zip file
> format, useful for trying it out but not for doing any development with it.
> Interest was so high that the eclipse.org web site was brought down by
> people trying to download the zips. Pushing to CVS was delayed until the
> team could talk about what directory and package names to use. But when a
> few small pieces were finally pushed to CVS, the package names were wrong
> anyway. :) That's Ok, things go wrong; it's not a big deal as long as you
> can iterate and fix them quickly. But now, according to the new milestone
> plan, it's hoped that there will be some editors working in 16 weeks, and
> the SQL viewer in 24 weeks.
>
> Meanwhile the talented developers at Genuitec, not burdened by any red tape
> or other bureaucratic issues, released a product using large parts of the
> IBM contribution on August 16. This is a good thing; they are *supposed* to
> productize the code, remember? But look, it only took them 25 days, counting
> weekends, from when the zip was available. This is even more impressive when
> you realize that to have a final, stable commercial GA product release on
> that date they had to have something compiling and running well before then.
> How long did that take, 5 days, 10 days?
>
> So on the one hand I see a large cumbersome open source project plodding
> along to do something in ~6 months what a small motivated group of agile
> programmers did in ~6 days. What is wrong with this picture?
>
> The developers who contributed all this good code must be incredibly
> frustrated to see this drag on for so long. I know I would be. Clearly the
> potential users are frustrated too. I apologize if this sounds overly
> critical but this project has been, and continues to be, an embarassment to
> Eclipse. It's so important, and there are so many talented people begging to
> work on this and contribute to it. So what is the problem?
>
> I don't care to speculate, but here is a simple plan to fix it. Deciding on
> package names and directory structures shouldn't take more than, say, a
> week, even if you're as long winded as I am. So one week from now, we would
> like to see every file from the IBM contribution, including any changes made
> in private IBM repositories since July, pushed to eclipse.org public
> webtools CVS. Builds should start immediately, by hand if necessary until
> the automated scripts can be set up. That's the first milestone. One week
> later, any changes made by others (such as Genuitec) since the initial
> contribution should be merged into public CVS. At this point there should be
> no private repositories, only the one true source at eclipse.org. Future
> milestones should be synced up to coincide with Eclipse Platform 3.1
> milestones and follow the excellent project management techniques pioneered
> by the Platform team. Code and concepts from the Lomboz contribution and
> anywhere else can be added on an ongoing basis. Simple.
>
> I challenge the current leadership of the webtools project to implement this
> plan immediately. Don't waste time explaining in great detail why it's not
> possible or whatever, just try it. Do I think it will happen? The cynic in
> me says no, but the optimist says maybe we'll be surprised. (Usually the
> cynic is right. :)
>
|
|
| |
| Re: Milestone Plan [message #47488 is a reply to message #47353] |
Fri, 01 October 2004 21:52 |
| Eclipse User |
|
|
|
While I agree with your goals I don't think this is a good way to achieve
them. Why? First, because too much of the work is done out of sight,
non-transparently. Second, by trying to build up from a nucleus you're
almost rewriting it from scratch. It's like saying, you don't like this
planet so let's blow it up and let gravity reform the pieces into a new one.
Finally, you're leaving a lot of code on the table unused. Like living
cells, code needs the oxygen of use and attention, and the waste removal of
maintenance and refactoring, or it will wither.
If you're worried about making too much API that's a valid concern. However,
refactoring is the typical way that code is reorganized into public and
private parts (i.e., external/internal). Standard practice for refactoring
is to a) have everything in source control, b) have unit tests, c) have the
system runable, nearly shippable at all times, d) make incremental,
mechanical changes, and e) repeat until done. Joel S. has a great essay on
this. Anyway, state up front that nothing is API until release, and API
should be requested with use cases and all that so there's a good reason for
it.
'nuf said :)
--
Ed
"Jochen Krause" <jkrause@w4toolkit.com> wrote in message
news:cjjl2j$9ls$1@eclipse.org...
Lets extract the smallest working
|
|
| |
| Re: Milestone Plan [message #47722 is a reply to message #47488] |
Mon, 04 October 2004 06:12 |
| Eclipse User |
|
|
|
Originally posted by: davegabol.eircom.net
"Ed Burnette" <ed.burnette@sas.com> wrote in message
news:cjl1av$jum$1@eclipse.org...
> While I agree with your goals I don't think this is a good way to achieve
> them. Why? First, because too much of the work is done out of sight,
(snip)
> private parts (i.e., external/internal). Standard practice for refactoring
> is to a) have everything in source control, b) have unit tests, c) have
the
> system runable, nearly shippable at all times, d) make incremental,
Hi Ed,
The examples you give are fine for the coding section of a project. But
there's much more to software engineering than coding as I'm sure you are
aware. These guys are trying to give us maintainable software not just
working software. The old saw about 60% of software cost being spent in
maintenance still holds true. Althought we won't be paying cash for
maintenance (adaptive, perfective etc) we will be paying in sweat, and
making others pay in broken builds every day unless this software is
architected properly. I agree it hasn't been the fastest start in the world
for the project, but to rush it now just to make up for that would be
criminal IMHO.
I don't have an answer to those who want/need something now, this very
second. Maybe an out of the way sandbox version of the original contrib
could be made so that people could learn the general shape of the code and
do a bit of prototyping of their own ideas? It would have to be clear
though, that nothing from this sandbox would be allowed into the eventual
"official" CVS repository unless it was based on the "official" code. So you
could play and code, but be prepared to make major mods if the base code you
used was thrown out.
Dunno, just a thought. I'd hate to see an unmaintainable pile of spaghetti
released just because the project team start feeling rushed.
All the best,
Dave.
|
|
| | | |
Goto Forum:
Current Time: Wed May 07 14:14:47 EDT 2025
Powered by FUDForum. Page generated in 0.04517 seconds |
Home