forrest-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Stefano Mazzocchi <stef...@apache.org>
Subject Re: import of xml.apache.org main site into forrest
Date Thu, 06 Jun 2002 20:22:01 GMT
Steven Noels wrote:
> 
> > From: Stefano Mazzocchi [mailto:stefano@apache.org]
> 
> > > I like this idea, we need to think about this.
> 
> 1) the 'we' was 'forrest', not outerthought

Ok, cool.

> > I have the impression that you guys might be caught in the good old
> > "don't who till is ready" ego trap with your 'libre' effort. Release
> > early and often means also 'release when it barely does anything' so
> 
> 2) touche :-(
> 
> point is that this thing has been sitting for quite some time on our
> harddisks (it's already more or less integrated in my local copy of
> forrest AAMOF), but since I said to my colleagues that I wanted to
> submit it to forrest, we have been discussing refactoring and some
> additional features ever since :-(

Yep, ego trap right there.

> I'm also aware however of the fact that providing a functional codebase
> helps with the adoption of new components, and we could go on endlessly
> in RT-mode if we don't come up with a codebase to start with. After all,
> Forrest has been mostly dead in the early days (sic) until Nicola and I
> jumpstarted it with the addition of the early centipede codebase, which
> was already pretty much cooked up prior to committing it

Exactly. Remember that evolution works best if done in small steps, not
in big jumps, because adaptability is maximum.

> > that people can interact with you *before* you spend months
> > developping
> > and find out that many things that you considered *very cool*
> > are simply
> > not needed anymore.
> 
> agree, we are anxiously trying to balance momentum and good design
> practices, and you know momentum is a prime mover & shaker in OSS
> projects. The result *will* be released under an Apache license however,
> whether it is used within Forrest or not, hopefully as being part of
> Forrest or Cocoon, or on SF. While the scope of libre is much smaller
> than the S&N portal contribution to Cocoon, its genesys should be
> comparable: it was mostly finished intra muros before being released as
> a contribution.

Ok
 
> > In fact, I don't want the book.xml to be removed by a
> > generator because
> > no matter how powerful the generator is the book.xml file can contain
> > much more metadata than any file system... and placing site-related
> > metadata (such the one I suggested above) inside the document and have
> > the generator extract it (a-la XPathDirectoryGenerator) seems like the
> > wrong approach to me since it mixes concerns between the document
> > authors and those who manage the site (which normally have different
> > concerns).
> 
> -1
> 
> I hate book.xml from a user's POV. It's non-intuitive, duplicates
> information, yet another thing people need to learn before they can
> trust project documentation to the Forrest service, etc...
> 
> For simple document collections, it should be sufficient to put files in
> a directory structure, and extract sequence and labeling from the
> filename or by applying XPath onto the files. To do so, we have a simple
> approach, configured using a little configuration file libre.xml, which
> you could compare with .htaccess
> 
> It's all under reconsideration right now, but what we have working
> currently is this:
> 
> libre.xml config file in content/xdocs:
> 
> <libre xmlns="http://outerx.org/libre/config/0.1">
>   <auto>
>     <filter logic="normal">
>       <property name="name" regex=".*\.xml"/>
>     </filter>
>     <label>
>       <xpath expression="/document/header/title/text()"/>
>     </label>
>   </auto>
> </libre>
> 
> Output of our libre generator:
> 
> <collection xmlns="http://outerx.org/yer/hierarchy/0.1">
>   <item/>
>   <item label="Contribution to Forrest"/>
>   <item label="The document-v1.1 DTD"/>
>   <item label="Forrest dream list"/>
>   <item/>
>   <item/>
>   <item label="Welcome to Forrest"/>
>   <item label="The Apache Software License, Version 1.1"/>
>   <item label="Mailing List Archives"/>
>   <item label="Mailing Lists"/>
>   <item label="The Forrest Primer"/>
>   <item label="Who we are"/>
> </collection>
> 
> As you can see, there are still issues (like directory handling in the
> above example, which we are addressing), but this would mean a lot of
> the information which now has to be hardcoded in book.xml can retrieved
> from the documents itself.
> 
> We have basically three actions we do on a hierarchy (and this stuff is
> being componentized so that we could implement this also on top of
> non-filesystem hierarchies like a WebDAV resource):
> 
>  - filtering
>  - labeling
>  - sorting
> 
> all three of them based on xpath expressions, resource properties
> (filename, mod date, etc) and regexes on these properties, with and'ing,
> or'ing and not'ing for filters.

Sounds very cool but...
 
> On your SoC remark: you're right, and maybe we will still need such an
> additional metadata config file for a filesystem repository, but WebDAV
> resource meta-attributes also move this responsibility to the individual
> resource maintainer, and not to the collection manager, if my
> assumptions are correct.

... I still question this.

> > I might be totally wrong and I'd love to be because I hate raining on
> > somebody else's party (I'm abusing this metaphor today) but your "*we*
> > need to think about this" rang a bell in my head.
> 
> you and me and other forresteers, thus, not 'we' ;-)

Ok, sorry for having misunderstood.

> <aside>
> > I still don't understand why we need to incorporate changes and todos
> > into a project description file.
> 
> and I can only agree with that.
> </aside>
> 
> Hopefully we find time to act upon our promises, you know we are trying
> really hard ;-)

Oh, gosh, I really don't care if people don't act upon their promises,
we are volunteers, damn it, there are tons of others things in our
priority list that come before this. You even have a new baby to take
care of. That's more than an excuse for me.

What I really dislike is the fact that people say their are going to do
something and they don't. But the problem is not that they didn't, is
that they told people!

For example: when I started Forrest, I never said I was working on
something, otherwise others (you and Nicola, in this case) wouldn't even
start any effort.

NOTE: I did this mistake several times before and this is why I'm
sensible about this.

So I changed about RT-style programming: tell people what the heck I
have in mind and tell them all my doubts, never telling them I'm working
on something. Work incrementally, commit something even if it doesn't
work as expected, even if it barely compiles. It's fine, it's what OSS
is about.

So, instead of telling us "we are thinking about redesign" and stopping
all our contributions for weeks, tell us your problems, give us the code
and we might even be able to solve them and release pressure and give
more time to you for other things.

Sure, you have to exit the ego trap to do this, but that's exactly the
obstacle that we must understand to reach the next step. ;-)

-- 
Stefano Mazzocchi      One must still have chaos in oneself to be
                          able to give birth to a dancing star.
<stefano@apache.org>                             Friedrich Nietzsche
--------------------------------------------------------------------



Mime
View raw message