tomcat-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Paulo Gaspar" <paulo.gas...@krankikom.de>
Subject RE: Collaborative Development (not Cat and Dog)
Date Tue, 15 May 2001 10:46:24 GMT
Hi Craig,


I am sorry my answer comes so late, but I am really overloaded at
the moment.

I sure understand your concerns and I sure would profit myself from
anything that makes Tomcat easier to install and use.

However, I can not agree with your conclusions on the nocive effects
of 3.3.

My evaluation of facts is:
 - There is almost no documentation on Tomcat, either on web pages
   or under a more formal format. Tomcat is much more complex than
   Velocity (a Jakarta project I know well) but its documentation
   looks pathetic when compared to Velocity's.
   IMO, this is the _obvious_ cause for that Tomcat users confusion
   that shows at the USER mailing list;

 - So, The effort on fighting 3.x as a source of confusion is
   misplaced. The constructive focus for that effort would be
   building better documentation about the Tomcat versions
   available, their differences and choice criteria. Common sense
   stuff like which JDK version and which Servlet/JSP versions each
   Tomcat version supports would be enough for the type of USER you
   refer;

 - Some of us really need a good enough Tomcat now. But:
     1. If development on 3.x had stopped, it would have been a big
        problem since Tomcat 3.2 still (now) has unsolved problems;
     2. Just counting on 4.x to solve this problem is hard because:
         * There is still no replacement for it, even because 4.0
	     depends on an external factor to become an official
           release (the release of the new JSP/Servlet standards);
         * 4.0 is still not acclaimed as being more stable than 3.x;
         * The lack of connectors for 4.x.

 - Many of those supporting 3.x did it because they needed that
   "good enough" Tomcat ASAP ("good enough" can have different
   meanings for each of us, of course). Mostly, their effort was
   never lost by Tomcat 4.x, because their focus was never there.
   Let's take a look at my self as a 3.x user: I am using Tomcat 3.3
   (even pre-release). If Tomcat 3.3 wasn't here, Tomcat 3.2 would
   be having even less attention and I would probably be using
   Resin;

 - Some of their effort is now being made available to Tomcat 4.x
   under several initiatives. Think about it:
    * _maybe_ Tomcat 4.x is now profiting from the efforts of
      people that would have moved away if 3.3 wasn't here.


My conclusions:
 1) Tomcat USER's bigger problem is lack of documentation, both on
    the project web pages and more formal documentation;

 2) Tomcat 3.3 may already be helping Tomcat 4.x to succeed and it
    will sure help us to have a better Tomcat 5.x.

I believe on 2) not because I think that 3.x will be the base of 5.x
but because I believe that 5.x will probably be based on the best
ideas from both 3.x and 4.x. And that is the way to go:
 - Giving space for different trends to be explored;
 - Using what proves to be better from all the available trends.

At least, this is the kind of software Darwinism I believe to be
usefulness. The other kind is based on egos and politics and I sure
hope that none of us (in this list) is interested on that one.
(Egos keep tempting us, but we must know better...)


Some further remarks:

>...
> If you want to understand why caring about users is important, go
> subscribe to TOMCAT-USER and start answering all of the plaintive calls
> for help.

Yes, I am a subscriber. I am not of any help (I tried some months ago)
just out of being overloaded. At the moment I am having to learn too much
too fast, both on APIs and OS products (which, of course, includes taking
a look at their source).

I only started using Java last Summer and am still catching up. My
background (with a lot of Delphi and C++) sure helps, but makes no
miracles.

Currently, I am a kind of Siamese twin of my PC.


> These people don't give a rip about the internal architecture -
> they are totally mystified by the absolutely horrible mess we've made of
> configuration, most especially for web connectors.  We're talking about
> lots of people who are new to servlets and JSP, and often new to Java --
> and we are erecting a huge wall in front of them, before they can
> productively use the software that we have been slaving over.
>
>...
>
> No argument that developers can do what they want in an open source
> project.  But, by MY definitions of the terms (since you're responding to
> what I said :-), there are orders of magnitude more Tomcat users in the
> world than there are Tomcat developers.  IMHO, we have an obligation to
> care about them as well, and not selfishly focus on us.

SURE. Config problems are the worst of it all.
SURE. We should care about those users. And what helps them helps all the
Tomcat users and developers.

But most of all, much better documentation is needed.

(Just don't look at me for that since:
  - I do not know enough;
  - my bigger itches are somewhere else;
  - I just have no time to be nice.
)

For me, the newer versions of 3.3 made my life MUCH easier by giving me
the option of a simpler CLASSPATH configuration.


>...
> By the way, I was an open source developer (see below) even before I
> joined Sun ... they hired me (at least in part) because of that.

I am aware of that. And I am sure you try hard to stick with Open Source
principles. (But Sun's interests pressure sure must be hard at times.)

>...
> > I am new here, but didn't Catalina start a bit against the flow?
> >
>
> Actually, it was the other way around if you look at the
> history.  Catalina was originally created in Apache JServ days -- it would
> have been Apache JServ 2.0, until Sun announced that they were
> contributing Tomcat in June 1999.  If you go peruse the JServ CVS tree on
> the JSERV1_1 branch, you'll see what is recognizably the same
> architecture and implementation.
>
> Catalina is actually going back to the roots of the Apache-based open
> source servlet container project :-).

Still, there was a time that Catalina was the revolution. It is so similar
to the 3.3 situation!

Anyway, I sure hope that the best of both solutions ends up being used.
There are very interesting solutions in each camp.

And I am not sure on how incompatible really are the "modules" ideas with
the "valves" ideas. Maybe the current implementations make that far apart,
but - from what I currently understand - I see advantages on both. Some
problems would be better solved with modules and some with valves.


>...
> > Hey! Wasn't Tomcat 4.x betting on mod_webapp???
> > You should be happy that a few 3.x USERS/Developers kept working on
> > mod_jk.
> >
> At the time that "bet" was placed, there were zero people working on
> mod_jk (and even less documentation :-) -- it wasn't an option that could
> be considered.
>
> It's still not clear to me that mod_jk will make it possible to create a
> combined webserver+Tomcat implementation that conforms to the specs --
> there's still a lot of work to be done there.

I understand both the motives of the mod_webapp bet and the current
problems.

But mod_jk is still the most stable base at the moment. There is a really
high probability that (perfect or not) it will be the solution used by
Tomcat 4.x for a while.

So, what I wrote remains true.


I was guessing that mod_webapp would become MUCH HARDER than expected. But
if I was on Pier shoes I would probably have gone for it anyway... and be
in trouble with it too. (I have been there.)

It is just that it is a kind of development with too much unforeseen. It
is the kind of thing that betrays someone that knows the subject even more
than someone that does not - the former tends to fall under the illusion
that he understands all the issues involved... until it is too late.
(I have been there too.)

It still has potential, but its potential usefulness depends a lot on being
a clean solution - which means that a lot of rewriting and refactoring
might happen.

I sure hope mod_webapp realizes its potential, but I also expect that:
 - mod_jk will be the way to go until that happens;
 - a lot of what is being learned on mod_jk can be used in mod_webapp.


>...
> You sparred with Jon and call *my* tone confrontational?  :-) :-) :-)

LOL
Of course that one can not compare!

Jon plays on a different division. An order of magnitude above!

(Just couldn't resist! But since Jon doesn't resist this kind of remarks
either... I hope he understands. =;o)
)


> Paulo, you've made some interesting and useful comments.  However,
> you're starting from a very different definition of USER than I am, and
> therefore drawing incorrect conclusions about what I am saying.

Maybe I tend to think of my self as just a USER of Tomcat. And maybe the
USERs you are talking about are not so similar to me as I expect.

On the conclusions, as I stated above, I understand about your motivations
to draw them but I still do not agree with them.


> I feel a personal, moral, obligation to the thousands and thousands of
> people who download Tomcat in the hopes of having a servlet container that
> works, and who want to have confidence in a solid support and future
> growth path -- and we (develoeprs) have not yet provided an unambiguous
> message to reassure them.

I believe that.

But I still think (as stated above) that:
 - Tomcats 3.x and 4.x are not contradictory but just complementary;
 - The main problem is the lack of documentation. (Just take a look at the
   Velocity pages and see the difference.)

I think that the "unambiguous message" just needs to be clearly stated on
the web pages.

Maybe the problem is that not much people believe that there is an
"unambiguous message" to be stated as much as I do. And that leads to a
waste of energies.

Selfish bastard as I am, I see that waste of energies as an obstacle to
better Tomcats 3.x, 4.x and 5.x. And since I want them (specially 5.x),
I keep trying to impose my own harmonious vision (a.k.a. "unambiguous
message") so that you all stop fighting each other and start working
together (even when trough different paths) for a brighter future.
=;o)


Ok, ok, ok! I promise I will start coding more and bulshiting less as soon
as I have time.

And yes... I can write little bits of code. I have a few witnesses
around, since I did a couple of those bits for Velocity...
Honest!!!
I swear!!!
=:o)


Have fun,

Paulo Gaspar



> -----Original Message-----
> From: Craig R. McClanahan [mailto:craigmcc@apache.org]
> Sent: Wednesday, May 09, 2001 9:01 PM
> To: tomcat-dev@jakarta.apache.org; paulo.gaspar@krankikom.de
> Subject: Re: Collaborative Development (not Cat and Dog)
>
>
>
>
> On Wed, 9 May 2001, Paulo Gaspar wrote:
>
> > > Besides developers and their own itches, there is another group whose
> > > interests ought to be considered - USERS of Tomcat.  If you don't have
> > > any, then it becomes a much less interesting project to work on.  And
> > > we've done a pretty good job at confusing people about what the Tomcat
> > > road map is.  If the user community were to abandon Tomcat,
> it wouldn't
> > > matter much what we do on the development side.
> >
> > Why are USERS that different from developers in this case?
> >
>
> By USERS, I'm talking about the people that use Tomcat as a product, and
> who don't usually even WANT to think about the Tomcat source code.  They
> don't include folks who want to think about participating in Tomcat's
> development -- they just want a cool servlet + JSP container.
>
> For a sense of perspective on the Tomcat user community, consider the
> following:
>
> Just off the main Apache web site, Tomcat binary downloads (in various
> release and nightly build formats) run from 50,000 to 100,000 downloads
> per month.  That doesn't count mirror sites (official and unofficial),
> people who download Tomcat as part of other app servers, people who
> download Tomcat as part of development tools packages, and people who
> download the J2EE RI (which includes Tomcat as its web layer).
>
> At the moment, there are 2,236 subscribers to the TOMCAT-USER list, which
> is the largest Jakarta mailing list other than ANNOUNCEMENTS (~4,200).
> TOMCAT-DEV has 1,125 - many of them subscribed because they want to
> understand where Tomcat is going (a pretty smart way to figure that out,
> IMHO), not because they are contributing.
>
> (This trend is visible in other Jakarta projects as well -- lots and lots
> of downloads, user mailing lists at least double the size of the developer
> list, well-supported packages gaining users much faster than they gain
> developers, ...)
>
> If you want to understand why caring about users is important, go
> subscribe to TOMCAT-USER and start answering all of the plaintive calls
> for help.  These people don't give a rip about the internal architecture -
> they are totally mystified by the absolutely horrible mess we've made of
> configuration, most especially for web connectors.  We're talking about
> lots of people who are new to servlets and JSP, and often new to Java --
> and we are erecting a huge wall in front of them, before they can
> productively use the software that we have been slaving over.
>
> > This is Open Source. The developers are USERS trying to scratch their
> > own itches.
> >
>
> Developers, in terms of the comments I was making, are often Users as well
> -- but there's a whole bunch more Users than that.  It's definitely open
> source, and we Developers are totally free to ignore the non-developer
> Users if we want to.  And those Users are totally free to ignore our
> software and flock to other products whose developers care about them.
> If we were to chase everyone away, it doesn't matter how cool our software
> is any more -- it becomes irrelevant.
>
> > For each USER that becomes a developer to scratch his own itch there are
> > several other USERS with the same itches just waiting for the scratching
> > solution to be ready. (*)
> >
> > One thing I had to learn while starting to use Open Source software (and
> > that was recently - during 2000) was:
> >  1) There is usually no fancy wrapped product with manual and you really
> >     have to get involved, subscribe the mailing lists and dive into the
> >     source code to understand some of the stuff you use;
> >  2) You still have (much) faster answers for your problems, even having
> >     the option of participating in their solution.
> >
> > Whomever is aware of 2) is motivated to do 1).
> > Whomever does 1) here, understands the 4.x versus 3.3 pros and cons and
> > is able to make the best choice for his own case.
> >
>
> No argument that developers can do what they want in an open source
> project.  But, by MY definitions of the terms (since you're responding to
> what I said :-), there are orders of magnitude more Tomcat users in the
> world than there are Tomcat developers.  IMHO, we have an obligation to
> care about them as well, and not selfishly focus on us.
>
> > Corporations need to have a black on white clear statement, hiding or
> > killing/loosing the advantages of Open Source's 'Software Darwinism'
> > (which seems to me to have a good profits/cost balance). But this is
> > not Sun's Tomcat Project, it is Apache's Jakarta Tomcat Project.
> >
>
> Corporations got that message a couple years ago :-)
>
> By the way, I was an open source developer (see below) even before I
> joined Sun ... they hired me (at least in part) because of that.
>
> > The others will always be the (prey) customers of the more traditional
> > commercial software or are better paying loads of money to have some
> > consultant thinking for them. (And the consultant might still be an
> > Open Source adept.)
> >
> >
> > > There isn't any compelling benefit (to the user community) to have two
> > > active development branches, once the "next generation" branch
> > > matures.  The 4.0 container code is pretty stable (letting us
> focus now on
> > > performance improvements and feature adds).
> >
> > You seem to imply that 4.0 is the final solution for everybody. But it
> > sure is not for me and for many others. (Keep in mind that I do NOT
> > think that 3.3 is the final solution for everybody either.)
> >
> > I am new here, but didn't Catalina start a bit against the flow?
> >
>
> Actually, it was the other way around if you look at the
> history.  Catalina was originally created in Apache JServ days -- it would
> have been Apache JServ 2.0, until Sun announced that they were
> contributing Tomcat in June 1999.  If you go peruse the JServ CVS tree on
> the JSERV1_1 branch, you'll see what is recognizably the same
> architecture and implementation.
>
> Catalina is actually going back to the roots of the Apache-based open
> source servlet container project :-).
>
> >
> > > performance improvements and feature adds).  We could have had mod_jk
> > > ported a long time ago if someone wanted to work on it then
> (as they are
> > > apparently willing to now).
> >
> > Hey! Wasn't Tomcat 4.x betting on mod_webapp???
> > You should be happy that a few 3.x USERS/Developers kept working on
> > mod_jk.
> >
>
> At the time that "bet" was placed, there were zero people working on
> mod_jk (and even less documentation :-) -- it wasn't an option that could
> be considered.
>
> It's still not clear to me that mod_jk will make it possible to create a
> combined webserver+Tomcat implementation that conforms to the specs --
> there's still a lot of work to be done there.
>
> >
> > Craig, I admire your work a lot. But I find it hard to understand why
> > such a brilliant person chooses a confrontational tone at the moment
> > he is collecting such benefits from cooperation.
> >
>
> You sparred with Jon and call *my* tone confrontational?  :-) :-) :-)
>
> Paulo, you've made some interesting and useful comments.  However,
> you're starting from a very different definition of USER than I am, and
> therefore drawing incorrect conclusions about what I am saying.
>
> I feel a personal, moral, obligation to the thousands and thousands of
> people who download Tomcat in the hopes of having a servlet container that
> works, and who want to have confidence in a solid support and future
> growth path -- and we (develoeprs) have not yet provided an unambiguous
> message to reassure them.
>
> > Remember: most 3.3 people were not that motivated to work on Catalina,
> > and because they worked (and keep working) on stuff for 3.3 that has
> > common usefulness, 4.x now can benefit form it.
> >
> > If it was not for 3.3, maybe they were working on something else. Are
> > you sure they would be converted to 4.x?
> >
> > And many USERS (like me) could be now investing on some other tool.
> > Staying around because of 3.3, maybe they can be converted to 4.x
> > AFTER 4.x proves its value.
> >
>
> Remember, by my definitions you're a DEVELOPER, not a USER :-).
>
> >
> > Have fun,
> > Paulo Gaspar
> >
> >
> > (*) - Most of them much more silent and even less participative than
> >       myself.  =;o)
> >
>
> Craig
>
>
>


Mime
View raw message