gump-general mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Stefano Mazzocchi <stef...@apache.org>
Subject Re: [RT] Was python a good idea?
Date Thu, 08 Jul 2004 20:22:42 GMT
Adam, please, let me start saying this is (as indicated) a random 
though, not a proposal, nor a criticism.

As Nicola said, moving from ant+xslt+bash to python was a tremendous 
improvement. I just wonder if we should stop there, especially given 
that this community is basically java gurus with a little big of python 
envy for some of the features (that could be replicated in java, if one 
*really* wanted them, btw).

Adam R. B. Jack wrote:

>>I have started to use python myself because I loved the much faster
>>try/fail cycle of a scripting language and python looked a lot
>>friendlier than other scripting languages.
> 
> Python is fun to get started with & has some really nice features. My guess
> is I've not even come close to touching the nicer parts ('cos I've yet to
> leave Java thinking far enough behind) and I kinda look forward to getting
> there, some day.
> 
>>But in my experience, it doesn't scale in terms of complexity as much as
>>java does.
> 
> Sure doesn't. 

Well, ask the Zope people about that: not many agree in python-guru land 
(and not only for ego or protection of their past decisions).

The strong vs. weak typing debate is vivid in my head.... and you can 
say that no matter how deep you go to describe the metadata associated 
to your dependencies, there is a point where all of them get weak.

Gump, for example, shows how the time variable makes a strongly typed 
system become weakly typed.

Also, if you get pretty heavy with Collections, even Java becomes weakly 
typed.

*but* for some reason, java scales better with size and internal 
complexity. it feels more solid, at least to me.

> Good practices (unit tests, getting good coverage, pychecker
> and all) can help, but any line/character not touched is a potential
> time-bomb. That said, those good practices are needed w/ any language, they
> just take discipline. 

Very true. Fact is this is a community of java programmers which a 
tendency for not really caring about religions and just getting the job 
done.

I'm not against python or your work, I just wonder if this isn't 
blocking all the tremendous java expertiese that we all have.

> Basically, I've come to live w/ that realization &
> stopped trying to take so many short cuts (despite knowing better). Maybe I
> owe Python thanks for that. ;-)
> 
> I've found a bunch of stuff I can't do w/ Python, but then I've found
> similar w/ Java. Python is clearly far less mature, but I doubt that will
> last very long. Basically, I have a love/hate relationship w/ Python (I've
> caused myself a lot of hours of pain), but I don't regret having tinkered
> with it.

I can sense that (otherwise, you have would have given up) but I also 
see that as much as Gump was a one-man-show (driven by Sam) this gump is 
another one man show (driven by you).

I consider myself Gump one of the most incredible innovations of the 
entire foundation. I wonder what's going to happen if you go away or 
stop being interested in that.

I'm not saying it's your fault, not at all... it's *our* fault that we 
don't contribute... but everytime I think about looking at the code the 
complexity of all that python just scares me away a little.... so I 
postpone and I never do it.

I *really* want to extend gump so that it builds apr httpd and 
subversion... but everytime I want to start, some other thing comes up 
and that itch goes away.

Yesterday was one of those days and I think I know realize that it's 
because of python.

I wonder how many others on this list feel the same, thus my RTs.

>>Also, it seems that there is a lot of black magic in getting it to run
>>very solidly, while java has years of polishing on seriously loaded
>>environments.
> 
> 
> That could easily be me. I chose to do some
> 
> 
>>So, I wonder: what would you think about a gump in java?
> 
> 
> Good idea? Maybe. Personally, I've invested so much in Python Gump to want
> to back off (although 6 months ago I'd've been there in a heart beat).
> Personally, I feel Gump is beautifully irrelevant -- if it dropped of the
> face of the planet  few people would notice -- but when it find issues &
> helps out, it is a sweet thing. 

Personally, I think gump is the most important project of the entire 
ASF, but it will take a few more years and a lot more work in certain 
other directions to show that.

> For me it is a beautify folly; so what
> better than to experiment with, to teach some Java folks a new tool, to
> learn Python & get a new perspective on the world?

If it floats your boat, I'm happy. I'm just concerned that the day you 
find something that is more interesting, we are left with a codebase 
that nobody knows how to maintain.

> BTW: Won't Java (with JDK 1.5 feature) + Groovy and Python be about the same
> thing sooner or later, anyway? ;-) The similarities will outweigh the
> differences. ;-)

There is a psycological impact in community dynamics that should not be 
underestimated.

> Basically, I think Python Gump was the right thing to do 'cos it breathed
> life into a somewhat mundane/infrastructural task. 

True.

> I do think it has become a barrier to entry for many, 
 > which I find disturbing.

Bingo.

> As such, I'd not fight
> against folks wanting to re-write in Java ('cos that is clearly quite
> doable).

Cool.

NOTE: I *DO NOT* have a plan to rewrite Gump in java anytime soon, I 
don't think I have the energy and itchness for it.

-- 
Stefano.


Mime
View raw message