Andrew Stevens wrote:
>On Sun, 2003-09-28 at 17:09, Aslak Helles=F8y wrote:
>>Since the start of the XDoclet2 project about a year ago I have=20
>>contributed 99% of all the code in XDoclet2. Solo man projects aren't=20
>>very popular at Jakarta. Further, Xdoclet2's nearest relatives=20
>>(PicoContainer, QDox and Middlegen) now live at Codehaus=20
>>So I propose to move the XDoclet2 core (along with Generama) to Codehau=
>>instead of Jakarta. That will also give me better control of the projec=
>>This proposed move to Codehaus does not concern the xdoclet-plugins=20
>>project at SourceForge.
>>What do you think?
>Other than being yet another site I'd have to register at, I don't
>really mind either way.
Whether XDoclet2 moves to Codehaus or Jakarta, you'll need to register a=20
new place anyway. If you intend to be a core committer, that is. Plugin=20
developers shouldn't have to care about the core.
> But what would we gain from it, other than the
>hassle of moving stuff across?=20
-On the technical side, full control over the server environment that=20
Codehaus offers, including:
o Working CVS
o Nightly Maven builds
o Continuous Integration
o ViewCVS with branch views
o Stable up-time
SF doesn't have any of these.
On the "brand" side, we get the Codehaus brand, which is definitaly up=20
and coming. There is a lot of high quality projects at Codehaus and more=20
Jakarta has a brand too, but lately many Jakarta committers have become=20
more and more frustrated with the politics and bureacracy there. -And=20
the quality of Jakarta projects is just a perception. The truth is that=20
there is a lot of crappy projects there now, and just a place where=20
people dump their projects to improve their market value.
> I thought you had admin rights on the SF
>project anyway, so what "better control" would it give? And what about
>the mailing lists, do we transfer them too (forcing everyone to
>re-subscribe) or leave them on here?
There are already new mailing lists for xdoclet2. Moving subscribers=20
from one place to another takes an admin 5 minutes.
>Moving to Jakarta I can understand; it has a reputation which makes its
>projects more readily accepted by risk-averse project managers and
>business types, which outweighs the hassle.
I'd rather host the project in a place that the committers can fully=20
control (instead of an external committee), and where it's possible to=20
have roles like "project manager". Jakarta imposes too many policies,=20
and is based on meritocracy (i.e. no project leads), and that is a model=20
I don't believe in anymore.
> Codehaus doesn't; it just
>makes seems like moving for the sake of moving. Feel free to convince
>me of the benefits, though :-) Perhaps their anonymous CVS performs
>better? (though SF do keep saying they're working on improving theirs)
SF is going down and under. Codehaus is up and coming. Jakarta is too=20
bureaucratic and not what it used to be.
Keep in mind that I'm talking about the XDoclet2/Generama core here, not=20
the plugins. The plugins will have their home at xdoclet-plugins at SF.
Anyway, the move of XDoclet2/Generama will only concern the committers,=20
which by so far is me and a couple of new people. (None of the other old=20
XDoclet committers have done anything significant on XDoclet2/Generama).
It's also important to note that the move to a new place has a symbolic=20
intent. I think we can all agree that XDoclet1.2.x is a project in very=20
bad shape. Code rot. Too many broken windows. Most developers are gone,=20
and the few that are left don't have the time or energy to fix the=20
endless stream of bugs (most of them without patches) to the test-less=20
hairball that XDoclet 1.2.x is. The reason why we get so few patches is=20
that it takes too much energy for average joe to figure out how the hell=20
to fix anything. It's sad to admit, but I agree with most of the things=20
Hani ranted about in his bileblog a week ago. Ara has responded to it in=20
The move of the new codebase to new places (XDoclet2/Generama to=20
Codehaus and the plugins to xdoclet-plugins@...) will underline that=20
XDoclet 1.2.x is dying, and hopefully make more developers and users=20
switch to XDoclet2, which is/has:
-based on better template engines (Velocity/Jelly)
-with a full test suite
-simple plugin development
-better build system (Maven)
-ide plugins (at least soon i hope)
>This sf.net email is sponsored by:ThinkGeek
>Welcome to geek heaven.
>xdoclet-devel mailing list