gump-general mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Sam Ruby <ru...@apache.org>
Subject Re: gump issues under winblows and a suggestion...
Date Mon, 13 Jan 2003 14:36:20 GMT
Nicola Ken Barozzi wrote:
> 
> If you searchg in the 
> alexandria-dev mailing list you will find some interesting discussions 
> about this. I'll try to summarize here the results I came to.

[snip]

> I refactored the Gump CVS to split the various parts of Gump into 
> repository data, aggregation of data, script generation, site 
> generation. Result? No real gain and nobody folowing. Deleted effort 

;-)

I can't resist:

         verba manent, scripta volant?

;-)

  = = =

Brief history of Gump.

Once upon a time, long long ago, I was the Tomcat 3.1 release manager. 
At that time, things were so wild and wooly, that in order to 
successfully build Tomcat on a daily basis one needed to make fixes to 
Ant.

Now, project had their own build scripts mostly based on Ant, so I 
simply created a script which ran them in series.

...

Over time, two things happened.  First, I got interested in more 
projects.  Second, changed the "echo" statements into my script that 
told me where it was at any point in time to echo HTML so that I could 
browse the results remotely.

The lovingly handcrafted script looked like this:

http://oss.software.ibm.com/developerworks/opensource/jakarta/buildall.html

and the output looked like this:

http://oss.software.ibm.com/developerworks/opensource/jakarta/buildlog.html

...

Still more time elapsed, and I found two problems.  One was that every 
time a project inserted a dependency, I had to figure out a correct 
order.  The second was that as the ASF grew, projects talked to each 
other less, so cross project breakages grew more commonplace.

The solution to the former was to automate the generation of the script. 
  The solution to the latter was to simply publish the results - 
ultimately augmeted by a nag script initially donated by Jon Stevens.

Why did I write it in XSLT?  For starters, Velocity didn't exist yet. 
And secondly, I wanted to learn XSLT.

...

Fast forward a bit more.  Gump certainly tried to follow Stefano's 
Mazzocchi's design pattern [1] of "good ideas and bad code build 
communities, the other three combinations do not", but the community 
never quite formed.  Why is that?  My theory is based Sunir's corrollary 
4b [2] "As long as the project looks like one person's work, it is one 
person's work.".  So, for a while now, I have tried to hold back.  Still 
not much luck, but I am still hopeful.

...

What does this all mean?  Don't be afraid that I'm going to -1 any 
alternative to XSLT generation of scripts.  Go for it.

...

Footnotes:
[1] 
http://archives.real-time.com/pipermail/cocoon-devel/2000-October/003023.html
[2] http://www.advogato.org/article/395.html#5

- Sam Ruby



Mime
View raw message