Mailing-List: contact gump-help@jakarta.apache.org; run by ezmlm
To: gump@jakarta.apache.org
From: Sam Ruby <rubys@apache.org>
Subject: Re: [RT] gump.py
Date: Mon, 14 Apr 2003 06:55:52 -0400
Lines: 77
Message-ID: <b7e42o$ffh$1@main.gmane.org>
References: <b7brbv$cq4$1@main.gmane.org> <b7bu6e$kik$1@main.gmane.org>
 <b7c00i$pit$1@main.gmane.org> <3E9A5F46.4040604@apache.org>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit
User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US;
 rv:1.3) Gecko/20030312
In-Reply-To: <3E9A5F46.4040604@apache.org>
Sender: news <news@main.gmane.org>

Nicola Ken Barozzi wrote:
> 
> Sam Ruby wrote, On 13/04/2003 17.34:
> ...
> 
>> This RT is not an experiment to see if I can code more efficiently in 
>> Python than Java, but to test the assertion that if Gump were recoded 
>> in Python that it would attract a larger community.
> 
> I have recently looked again at Gump. IMVHO the problem with Gump ATM is 
> not the language it's made with.
> 
> Jenny merges the stuff nicely, and works for me. Whatever it does, it 
> merges stuff and produces xml, that is quite understandable by all 
> coders, be them feom java, perl, C, or whatever.

The DOM code is a bit challenging.

> The problem I have is with what happens after that. Just quick 
> observations:
> 
> 1) Some xslts generate scripts that write out xml, and this is *very* 
> twisted and confusing.

My guess is there wouldn't be any computer authored scripts in a Python 
version.

> 2) I don't understand the scripts and what they do. Probably comments 
> would help.

My guess is there wouldn't be any computer authored scripts in a Python 
version.

> 3) Having gen script *and* Ant buildfile is also confusing. IMHO we 
> should ditch one of the two. IMHO the scripts, and Ant is 
> cross-platform, so we need only one version ;-)

Most people launch ant from a script: build.bat, build.sh, ant.bat, or 
the like.

When I started at Apache, I often had to make fixes to Ant to get Tomcat 
to build.  Most ant.bat files at that time simply did things like set up 
the classpath and then invoke java org.apache.tools.ant... I started by 
calling these scripts from a script of my own, but eventually it was 
easier to simply set the classpath myself and invoke Ant directly.  As 
the dependencies grew, it made sense to let the computer maintain them. 
  So, I invented an XML descriptor and generated them.

I experimented several times with the conditional logic in Ant and 
writing out of the HTML from Ant.  Each time, the end result was faster 
and easier to understand in batch files and bash scripts.

Unquestionably, Python represents an improvement over this.

> 4) The fact that the scripts are generated is also confusing because the 
> result is not commented and very very big. One does not follow the flow, 
> moreover of scripts he never really got to write.

My guess is there wouldn't be any computer authored scripts in a Python 
version.

> 5) If something goes wrong in the script, it's hard to see why. This 
> needs better error reporting, and better reporting overall.

This is something that will be *much* easier in Python than batch or 
bash scripts.  Or in Ant scripts for that matter.

> All this brings me to think that before recoding in Python, which could 
> be better, I dunno, maybe we should get to make the above better.
> 
> Still diggin...

Summary: no computer authored scripts.  Better and more meaningful error 
messages.

- Sam Ruby