You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@ant.apache.org by Ara Abrahamian <ar...@yahoo.com> on 2002/05/19 20:11:43 UTC
RE: [VOTE] target-less build files (plus a new proposal)
It's like a book without chapters :-) Boring, non-modular.
With targets you have a simple mechanism to define small functions in
your build, call them, depend on them. Like you "extract variable" in
refactoring of your java code, you also "extract target" in your build
file, you give it a good name and the build file is more readable then.
I really don't find a target-less build file useful. Ant's build files
are generally hard to read and target names help a lot to understand
what's going on without digging into individual task usages. Some target
names like "main" are generally used everywhere as the base target. And
imho, like Java language itself, we should encourage ppl to use good
practices and even disallow bad practices, or leave them out of the
system at least, like Java.
Why should you use a target-less build.xml? Why? What is it you think
it's useful? :-)
Btw, I'm not sure we got into any near-term conclusion about reusable
build files. Seems like Myrmidon has some templating functionality, but
nothing for the 1.x line. What I really like about Tapestry web app
framework is how it handles reusable components. Say I want to define a
reusable ContactList component which shows a set of names/phones in
table rows/columns in my dynamic html page and use it in different
pages. I do it like this:
I define a ContactList.html file, there I define the html look of the
component (note that in jsp you define it in java code):
<table class="BigFont">
<span jwcid="forAllContactRows">
<tr valign=top>
<td>
<span jwcid="insertName">Name goes here.</span>
</td>
...
</tr>
</span>
</table>
Then there is a jwc file, I define the caller interface of the
component, plus the fill in the blank parts (marked with jwcid
attributes):
<specification
class="com.primix.tapestry.BaseComponent"
allow-informal-parameters="no" allow-body="yes">
<parameter name="contactRows" java-type="java.util.List"
required="yes"/>
<component id="forAllContactRows" type="Foreach">
<binding name="source" property-path="contactRows"/>
</component>
...
</specification>
So there's this <parameter/> element, the user knows that it should pass
a List to the component, and forAllContactRows and other jwcids are
actually defined here (forAllContactRows uses Foreach component of
Tapestry, for example). Ant tasks are like jsp taglibs, a code is behind
the task, not a piece of a build file. Note that if I want to do some
code-behind-component, I just change the class attribute to point to my
class, and put my sophisticated component logic there.
What's got this to do with Ant? Well, think of ContactList.html a piece
of a build file. I use <jar/>/<javac/> and other tasks of Ant. Now
forget about the <component/> in the ContactList.jwc file, but
concentrate on <parameter/>. I can define input parameters for my build
components. It's waaaaaaay cleaner than the current approach of Ant: ant
properties.
So what I'm proposing is a single .arc file (Ant Reusable Component!),
with a <component/> root element and this root element has some
<parameter/> elements which defines the parameters the ARC accepts (say
the compileDir/srcDir/etc for a reusable ARC which does the compiling
steps for me). Now whenever I want to define a reusable ant component I
just create an .arc file, define the parameters there and use it like
this:
<target name="x">
<component-alias name="compile" path="./compile.arc"/>
<compile>
<compileDir>../output</compileDir>
...
</compile>
...
</target>
So now I have an easy and modular way of defining build pieces, reusable
build scripts. Note that it let's you create components but the need for
overriding or calling or depending to targets in this or other build
files still exists and I think a fair amount of discussion went on
recently on this subject (but I'm not sure about the end result!). Imho
this <parameter/> mechanism should be used in definition of targets too.
Some targets have depends="", some don't or shouldn't because they
should only be called with an antcall. So for those only callable with
an antcall we can define a clean interface by using <parameter/>
elements in their definition. Relying solely on globally defined
<property/> elements is bad. An added benefit of this system is that we
can probably pipeline outputs of one build piece to the other build
pieces (or maybe call it buildlets?! I like this name!). Get a list of
files from one component as an out parameter, flush it into another one
as an in parameter.
So what do you think?
Ara.
> -----Original Message-----
> From: Diane Holt [mailto:holtdl@yahoo.com]
> Sent: Sunday, July 21, 2002 3:05 AM
> To: Ant Developers List
> Subject: RE: [VOTE] target-less build files
>
> --- Ara Abrahamian <ar...@yahoo.com> wrote:
> > - A build without any targets is a bad practice, isn't it?
>
> Why? What is it you think will happen?
>
> Diane
>
> =====
> (holtdl@yahoo.com)
>
>
>
> __________________________________________________
> Do You Yahoo!?
> Yahoo! Health - Feel better, live better
> http://health.yahoo.com
>
> --
> To unsubscribe, e-mail:
<ma...@jakarta.apache.org>
> For additional commands, e-mail:
<ma...@jakarta.apache.org>
--
To unsubscribe, e-mail: <ma...@jakarta.apache.org>
For additional commands, e-mail: <ma...@jakarta.apache.org>
Re: Vetoes are void? was Re: [VOTE] target-less build files -
countingresults
Posted by "Patrick (Gus) Heck" <pa...@olin.edu>.
Well you are right, though in the specific case I was refering to
nothing was final. I tend to like to see such constants attached to a
class that makes sense rather than collected in a constant bag. I'd also
consider creating singlton objects that held the resource...
I was also thinking of the more extreme case where a person actually
writes a bunch of non obj oriented code that does non-intializer type
work and gets run at startup before main. I'm sure someone out there is
doing that. There might even be a good reason for it... But I wouldn't
call it a good or clean use of java. IMHO it would be a hack that was
working around limitations of java if it was neccessary and just crappy
code if it wasn't. At least that is my opinion.
Gus
"J.Pietschmann" wrote:
>
> Patrick (Gus) Heck wrote:
> >>>...is like having methodless classes with only
> >>>a static part...
> > In fact a colleague of mine just ran into exactly this type of java file
> > (written by someone else) in some code he is working with. Scary but
> > there are people out there who do scary things!
>
> What's scary about them? In some circumstances Java
> classes with only final static variables are a perfect
> place for keeping magic constants, or hash tables (this
> is where static initializers come into play).
> I've seen this regularly in context of generated code.
>
> In the same sense I could imagine build files wich
> only define properties, to be imported. But then,
> it's a matter of taste whether you like them or not.
>
> J.Pietschmann
>
> --
> To unsubscribe, e-mail: <ma...@jakarta.apache.org>
> For additional commands, e-mail: <ma...@jakarta.apache.org>
--
To unsubscribe, e-mail: <ma...@jakarta.apache.org>
For additional commands, e-mail: <ma...@jakarta.apache.org>
Re: Vetoes are void? was Re: [VOTE] target-less build files - counting
results
Posted by "J.Pietschmann" <j3...@yahoo.de>.
Patrick (Gus) Heck wrote:
>>>...is like having methodless classes with only
>>>a static part...
> In fact a colleague of mine just ran into exactly this type of java file
> (written by someone else) in some code he is working with. Scary but
> there are people out there who do scary things!
What's scary about them? In some circumstances Java
classes with only final static variables are a perfect
place for keeping magic constants, or hash tables (this
is where static initializers come into play).
I've seen this regularly in context of generated code.
In the same sense I could imagine build files wich
only define properties, to be imported. But then,
it's a matter of taste whether you like them or not.
J.Pietschmann
--
To unsubscribe, e-mail: <ma...@jakarta.apache.org>
For additional commands, e-mail: <ma...@jakarta.apache.org>
Re: Vetoes are void? was Re: [VOTE] target-less build files - counting results
Posted by Stephane Bailliez <sb...@apache.org>.
----- Original Message -----
From: "Patrick (Gus) Heck" <pa...@olin.edu>
>
> In fact a colleague of mine just ran into exactly this type of java file
> (written by someone else) in some code he is working with. Scary but
> there are people out there who do scary things!
There is nothing scary, you can have a complex initialization part of the
static block like you could compute a table, precompile regex or whatever.
The huge difference is this case is as mentionned by Nikola that the code is
structured and inside a static block it's not everywhere between the methods
with no block nor anything.
If there is a target less build file I would be in favor of forcing order
and put all elements before <target>, but then again this would break
backward compatibility. I can't even think of an initialization stuff that
will be triggered at the startup but spreaded all over the build file: top ,
in between, bottom. Food for nightmare build file but then again it might be
a still issue and the one writing this should be shot.
Stephane
--
To unsubscribe, e-mail: <ma...@jakarta.apache.org>
For additional commands, e-mail: <ma...@jakarta.apache.org>
Re: Vetoes are void? was Re: [VOTE] target-less build files - counting
results
Posted by "Patrick (Gus) Heck" <pa...@olin.edu>.
Magesh Umasankar wrote:
>
> ----- Original Message -----
> From: "Nicola Ken Barozzi" <ni...@apache.org>
> >
> > Listen guys, if we follow the analogy with Java classes, what Peter says
> > make perfect sense.
>
> > Having targetless buildfiles is like having methodless classes with only
> > a static part... I don't know if it's possible in java, but who would
> > ever use it?
>
> It is very much possible in Java to have just
> a static part and nothing else.
In fact a colleague of mine just ran into exactly this type of java file
(written by someone else) in some code he is working with. Scary but
there are people out there who do scary things!
Gus
--
To unsubscribe, e-mail: <ma...@jakarta.apache.org>
For additional commands, e-mail: <ma...@jakarta.apache.org>
Re: Vetoes are void? was Re: [VOTE] target-less build files - counting results
Posted by Magesh Umasankar <um...@apache.org>.
----- Original Message -----
From: "Nicola Ken Barozzi" <ni...@apache.org>
>
> Listen guys, if we follow the analogy with Java classes, what Peter says
> make perfect sense.
> Having targetless buildfiles is like having methodless classes with only
> a static part... I don't know if it's possible in java, but who would
> ever use it?
It is very much possible in Java to have just
a static part and nothing else. Just because
it doesn't have a "main" method doesn't mean
that the class's static initializers cannot
be run. In fact you don't even need to
instantiate the class for its static initializers
to run. The default class loader will run the
static initializers if you just do Class.forName().
Java doesn't mandate that every class must have
a main method. Java does not mandate that every
class must have at least a method.
I fail to understand why we are trying
to draw an analogy between Java - a full
fledged language - and Ant. If we start going
that route, one can start drawing analogies
between numerous other things like constants
and variables in Java, anonymous classes,
package names... Well, you get the idea ;-)
Cheers,
Magesh
*************************************************
* Committee: Individuals who can do nothing *
* individually and sit to decide that nothing *
* can be done together. *
*************************************************
--
To unsubscribe, e-mail: <ma...@jakarta.apache.org>
For additional commands, e-mail: <ma...@jakarta.apache.org>
Re: Vetoes are void? was Re: [VOTE] target-less build files - counting results
Posted by Stephane Bailliez <sb...@apache.org>.
----- Original Message -----
From: "Nicola Ken Barozzi" <ni...@apache.org>
>
> Listen guys, if we follow the analogy with Java classes, what Peter says
> make perfect sense.
> Having targetless buildfiles is like having methodless classes with only
> a static part... I don't know if it's possible in java, but who would
> ever use it?
Sorry, I own the Trademark for this analogy ;-)
http://marc.theaimsgroup.com/?l=ant-dev&m=102698192030207&w=2
Stephane
--
To unsubscribe, e-mail: <ma...@jakarta.apache.org>
For additional commands, e-mail: <ma...@jakarta.apache.org>
Re: Vetoes are void? was Re: [VOTE] target-less build files -
counting results
Posted by Martin van den Bemt <ml...@mvdb.net>.
It still has methods from Object...
Mvgr,
Martin
On Wed, 2002-07-24 at 19:43, Jose Alberto Fernandez wrote:
> --- Nicola Ken Barozzi <ni...@apache.org> wrote:
> >
> >
> > Listen guys, if we follow the analogy with Java
> > classes, what Peter says
> > make perfect sense.
> >
> > Having targetless buildfiles is like having
> > methodless classes with only
> > a static part... I don't know if it's possible in
> > java, but who would
> > ever use it?
> >
>
> public class Constants {
> public static final Wigget THE_WIGET;
> public static final Googedly GOOK;
>
> static {
> try {
> THE_WIGET = new Wigget();
> GOOK = new Googedly();
> }
> catch (Exception e) {
> System.out.println("Could not initialize");
> }
> }
> }
>
> Actually this is a quite useful pattern, which I have
> seen in many places.
>
> Jose Alberto
>
>
> __________________________________________________
> Do You Yahoo!?
> Everything you'll ever need on one web page
> from News and Sport to Email and Music Charts
> http://uk.my.yahoo.com
>
> --
> To unsubscribe, e-mail: <ma...@jakarta.apache.org>
> For additional commands, e-mail: <ma...@jakarta.apache.org>
>
>
--
To unsubscribe, e-mail: <ma...@jakarta.apache.org>
For additional commands, e-mail: <ma...@jakarta.apache.org>
Re: Vetoes are void? was Re: [VOTE] target-less build files - counting results
Posted by Jose Alberto Fernandez <j_...@yahoo.com>.
--- Nicola Ken Barozzi <ni...@apache.org> wrote:
>
>
> Listen guys, if we follow the analogy with Java
> classes, what Peter says
> make perfect sense.
>
> Having targetless buildfiles is like having
> methodless classes with only
> a static part... I don't know if it's possible in
> java, but who would
> ever use it?
>
public class Constants {
public static final Wigget THE_WIGET;
public static final Googedly GOOK;
static {
try {
THE_WIGET = new Wigget();
GOOK = new Googedly();
}
catch (Exception e) {
System.out.println("Could not initialize");
}
}
}
Actually this is a quite useful pattern, which I have
seen in many places.
Jose Alberto
__________________________________________________
Do You Yahoo!?
Everything you'll ever need on one web page
from News and Sport to Email and Music Charts
http://uk.my.yahoo.com
--
To unsubscribe, e-mail: <ma...@jakarta.apache.org>
For additional commands, e-mail: <ma...@jakarta.apache.org>
Re: Vetoes are void? was Re: [VOTE] target-less build files - counting
results
Posted by co...@covalent.net.
On Wed, 24 Jul 2002, Nicola Ken Barozzi wrote:
> Listen guys, if we follow the analogy with Java classes, what Peter says
> make perfect sense.
>
> Having targetless buildfiles is like having methodless classes with only
> a static part... I don't know if it's possible in java, but who would
> ever use it?
Nicola, this is not about 'if what Peter says makes sense' - it
is about what the majority of ant commiters thinks.
And it is about how decisions are made - by one person (or 2) or
by majority.
You have a very valid point with <init>, and if a majority of
ant commiters share this view, that'll be how ant will work,
even if I don't agree and vote -1.
I voted -1 on removing tasks from the top level, yet a majority
at that time decided to remove them and they were removed.
Now, based on what we've learned - we have a pretty strong majority
who things tasks should be allowed - and again that's what should
happen even if few people disagree.
Please make distinction about the validity of a point and the
decision-making process we should follow.
Costin
--
To unsubscribe, e-mail: <ma...@jakarta.apache.org>
For additional commands, e-mail: <ma...@jakarta.apache.org>
Re: Vetoes are void? was Re: [VOTE] target-less build files - counting
results
Posted by Nicola Ken Barozzi <ni...@apache.org>.
Peter Donald wrote:
> On Wed, 24 Jul 2002 20:19, Stefan Bodewig wrote:
>
>>On Wed, 24 Jul 2002, Peter Donald <pe...@apache.org> wrote:
>>
>>>So far I have have vetoed the change and given reasons which other
>>>people have supported.
>>
>>I've cited all reasons as far as I can tell. These reasons are
>>reasons against top-level tasks, not against target-less builds.
>
>
> I have no problem with top level tasks and I have never said I did as far as I
> recall. I would prefer that only declarators were part of top level but as
> that is nearly impossible to implement due to ants wonderful codebase I have
> no problem with making any task a top level task.
>
>
>>You can not honestly claim that my example (b) is cleaner, clearer,
>>easier to understand or makes the learning curve smoother than my
>>example (a).
>
>
> I can claim that
>
> <project>
> <target name="main">
> <echo>Hello world</echo>
> </target>
> </project>
>
> is cleaner, clearer, easier to understand ... yada yada.
>
> It follows our current model where work is described by targets and is
> requires very little effort to migrate into a fully fledged, well modularised
> build file.
>
> Where I don't think targetless build files do the same. Stephane seems to
> agree with me thus my -1 is not invalid.
Listen guys, if we follow the analogy with Java classes, what Peter says
make perfect sense.
Having targetless buildfiles is like having methodless classes with only
a static part... I don't know if it's possible in java, but who would
ever use it?
--
Nicola Ken Barozzi nicolaken@apache.org
- verba volant, scripta manent -
(discussions get forgotten, just code remains)
---------------------------------------------------------------------
--
To unsubscribe, e-mail: <ma...@jakarta.apache.org>
For additional commands, e-mail: <ma...@jakarta.apache.org>
Re: Vetoes are void? was Re: [VOTE] target-less build files - counting results
Posted by Stefan Bodewig <bo...@apache.org>.
On Thu, 25 Jul 2002, Peter Donald <pe...@apache.org> wrote:
> I have no problem with top level tasks and I have never said I did
> as far as I recall.
You do, because saying
> <project>
> <target name="main">
> <echo>Hello world</echo>
> </target>
> </project>
is cleaner than
<project>
<echo>Hello world</echo>
</project>
says <echo> should be inside a <target>.
And the question whether we allow build files without any targets at
all really boils down to my two (extreme, granted) examples.
> I can claim that
>
> <project>
> <target name="main">
> <echo>Hello world</echo>
> </target>
> </project>
>
> is cleaner, clearer, easier to understand
I don't think it easier to understand. I imediately agree it makes
learning Ant's target concept easier, once you've written the above.
> ... yada yada.
I was citing you 8-)
> Where I don't think targetless build files do the same.
They don't. But they are appropriate for cases that don't need to be
modularized at all (and will never be).
Stefan
--
To unsubscribe, e-mail: <ma...@jakarta.apache.org>
For additional commands, e-mail: <ma...@jakarta.apache.org>
Re: Vetoes are void? was Re: [VOTE] target-less build files - counting results
Posted by Peter Donald <pe...@apache.org>.
On Thu, 25 Jul 2002 03:37, Jose Alberto Fernandez wrote:
> > It follows our current model where work is described
> > by targets and is
> > requires very little effort to migrate into a fully
> > fledged, well modularised
> > build file.
>
> Why is it so much more dificult to add:
> <target name="main">
> </target>
>
> Adding those two lines around a target-less list of
> actions does not seem too more dificult to me.
> Is there some specific horrible escenario that you are
> concern about?
I think you will find I am advocating the addition of
<target name="main">
</target>
--
Cheers,
Peter Donald
--------------------------------
My opinions may have changed,
but not the fact that I am right
--------------------------------
--
To unsubscribe, e-mail: <ma...@jakarta.apache.org>
For additional commands, e-mail: <ma...@jakarta.apache.org>
Re: Vetoes are void? was Re: [VOTE] target-less build files - counting results
Posted by Stephane Bailliez <sb...@apache.org>.
----- Original Message -----
From: <co...@covalent.net>
[...]
> All I want is to have an end to the interminable discussions
> where we all repeat the same arguments - and a majority
> vote should be able to stop it ( at least for a while ).
This discussion is closed for me and target less build files are in as it
has been accepted by the majority.
Period.
--
To unsubscribe, e-mail: <ma...@jakarta.apache.org>
For additional commands, e-mail: <ma...@jakarta.apache.org>
Re: Vetoes are void? was Re: [VOTE] target-less build files - counting
results
Posted by co...@covalent.net.
On Wed, 24 Jul 2002, Jose Alberto Fernandez wrote:
> Costin, calm down. I am on your side. :-)
:-)
It's just that I've heard the word 'veto' too much over
the last few days...
Since you mention this - I also have a problem
with 'sides' :-) There are too many people with strong opinions here,
and 'sides' are a dangerous concept.
All I want is to have an end to the interminable discussions
where we all repeat the same arguments - and a majority
vote should be able to stop it ( at least for a while ).
Costin
>
> --- costinm@covalent.net wrote: > On Wed, 24 Jul
> 2002, Jose Alberto Fernandez wrote:
> >
> > > Actually, there even was an implementation of this
> > > declarators, which provided exactly this, but was
> > > vetoed (I forget by whom). It worked quite well,
> > we
> > > can bring it back if people like. ;-)
> >
> > Make a formal proposal, put it to vote - and if it
> > gets
> > a majority of votes it can be implemented. The
> > implementation
> > can be vetoed, the proposal is a majority vote.
> >
> > We do have a majority of votes on top-level tasks
> > and
> > targetless tasks - but that doesn't mean <init> or
> > any other proposal that gets majority can't be
> > implemented
> > and used.
> >
> > I'm -0 on <init> and I'll probably not vote against
> > any other
> > style. If 3 commiters want to use a certain style
> > and that
> > doesn't affect me - I see no reason to vote against.
> >
> >
> > > Why is it so much more dificult to add:
> > > <target name="main">
> > > </target>
> > >
> > > Adding those two lines around a target-less list
> > of
> > > actions does not seem too more dificult to me.
> > > Is there some specific horrible escenario that you
> > are
> > > concern about?
> >
> > There is nothing wrong with doing that, and nothing
> > stops
> > you ( or anyone ) from doing exactly that.
> >
> > If you think that this should be the only solution -
> > put
> > this to a vote, get the majority to revert the vote
> > on
> > targetless - and we'll live with that.
> >
> > We can fight forever over taste and preferences over
> > a
> > style or another - and I think we should be as open
> > as
> > possible with other's preference. Even if something
> > is turned down by majority, if enough commiters want
> > something we should at least provide some hooks to
> > allow them to do that outside of ant.
> >
> >
> > Costin
> >
> >
> > --
> > To unsubscribe, e-mail:
> > <ma...@jakarta.apache.org>
> > For additional commands, e-mail:
> > <ma...@jakarta.apache.org>
> >
>
> __________________________________________________
> Do You Yahoo!?
> Everything you'll ever need on one web page
> from News and Sport to Email and Music Charts
> http://uk.my.yahoo.com
>
> --
> To unsubscribe, e-mail: <ma...@jakarta.apache.org>
> For additional commands, e-mail: <ma...@jakarta.apache.org>
>
>
--
To unsubscribe, e-mail: <ma...@jakarta.apache.org>
For additional commands, e-mail: <ma...@jakarta.apache.org>
Re: Vetoes are void? was Re: [VOTE] target-less build files - counting results
Posted by Jose Alberto Fernandez <j_...@yahoo.com>.
Constin, calm down. I am on your side. :-)
--- costinm@covalent.net wrote: > On Wed, 24 Jul
2002, Jose Alberto Fernandez wrote:
>
> > Actually, there even was an implementation of this
> > declarators, which provided exactly this, but was
> > vetoed (I forget by whom). It worked quite well,
> we
> > can bring it back if people like. ;-)
>
> Make a formal proposal, put it to vote - and if it
> gets
> a majority of votes it can be implemented. The
> implementation
> can be vetoed, the proposal is a majority vote.
>
> We do have a majority of votes on top-level tasks
> and
> targetless tasks - but that doesn't mean <init> or
> any other proposal that gets majority can't be
> implemented
> and used.
>
> I'm -0 on <init> and I'll probably not vote against
> any other
> style. If 3 commiters want to use a certain style
> and that
> doesn't affect me - I see no reason to vote against.
>
>
> > Why is it so much more dificult to add:
> > <target name="main">
> > </target>
> >
> > Adding those two lines around a target-less list
> of
> > actions does not seem too more dificult to me.
> > Is there some specific horrible escenario that you
> are
> > concern about?
>
> There is nothing wrong with doing that, and nothing
> stops
> you ( or anyone ) from doing exactly that.
>
> If you think that this should be the only solution -
> put
> this to a vote, get the majority to revert the vote
> on
> targetless - and we'll live with that.
>
> We can fight forever over taste and preferences over
> a
> style or another - and I think we should be as open
> as
> possible with other's preference. Even if something
> is turned down by majority, if enough commiters want
> something we should at least provide some hooks to
> allow them to do that outside of ant.
>
>
> Costin
>
>
> --
> To unsubscribe, e-mail:
> <ma...@jakarta.apache.org>
> For additional commands, e-mail:
> <ma...@jakarta.apache.org>
>
__________________________________________________
Do You Yahoo!?
Everything you'll ever need on one web page
from News and Sport to Email and Music Charts
http://uk.my.yahoo.com
--
To unsubscribe, e-mail: <ma...@jakarta.apache.org>
For additional commands, e-mail: <ma...@jakarta.apache.org>
Re: Vetoes are void? was Re: [VOTE] target-less build files - counting
results
Posted by co...@covalent.net.
On Wed, 24 Jul 2002, Jose Alberto Fernandez wrote:
> Actually, there even was an implementation of this
> declarators, which provided exactly this, but was
> vetoed (I forget by whom). It worked quite well, we
> can bring it back if people like. ;-)
Make a formal proposal, put it to vote - and if it gets
a majority of votes it can be implemented. The implementation
can be vetoed, the proposal is a majority vote.
We do have a majority of votes on top-level tasks and
targetless tasks - but that doesn't mean <init> or
any other proposal that gets majority can't be implemented
and used.
I'm -0 on <init> and I'll probably not vote against any other
style. If 3 commiters want to use a certain style and that
doesn't affect me - I see no reason to vote against.
> Why is it so much more dificult to add:
> <target name="main">
> </target>
>
> Adding those two lines around a target-less list of
> actions does not seem too more dificult to me.
> Is there some specific horrible escenario that you are
> concern about?
There is nothing wrong with doing that, and nothing stops
you ( or anyone ) from doing exactly that.
If you think that this should be the only solution - put
this to a vote, get the majority to revert the vote on
targetless - and we'll live with that.
We can fight forever over taste and preferences over a
style or another - and I think we should be as open as
possible with other's preference. Even if something
is turned down by majority, if enough commiters want
something we should at least provide some hooks to
allow them to do that outside of ant.
Costin
--
To unsubscribe, e-mail: <ma...@jakarta.apache.org>
For additional commands, e-mail: <ma...@jakarta.apache.org>
Re: Vetoes are void? was Re: [VOTE] target-less build files - counting results
Posted by Jose Alberto Fernandez <j_...@yahoo.com>.
--- Peter Donald <pe...@apache.org> wrote: > On Wed,
24 Jul 2002 20:19, Stefan Bodewig wrote:
>
> I have no problem with top level tasks and I have
> never said I did as far as I
> recall. I would prefer that only declarators were
> part of top level but as
> that is nearly impossible to implement due to ants
> wonderful codebase I have
> no problem with making any task a top level task.
>
Actually, there even was an implementation of this
declarators, which provided exactly this, but was
vetoed (I forget by whom). It worked quite well, we
can bring it back if people like. ;-)
> It follows our current model where work is described
> by targets and is
> requires very little effort to migrate into a fully
> fledged, well modularised
> build file.
>
Why is it so much more dificult to add:
<target name="main">
</target>
Adding those two lines around a target-less list of
actions does not seem too more dificult to me.
Is there some specific horrible escenario that you are
concern about?
Jose Alberto
__________________________________________________
Do You Yahoo!?
Everything you'll ever need on one web page
from News and Sport to Email and Music Charts
http://uk.my.yahoo.com
--
To unsubscribe, e-mail: <ma...@jakarta.apache.org>
For additional commands, e-mail: <ma...@jakarta.apache.org>
Re: Vetoes are void? was Re: [VOTE] target-less build files - counting
results
Posted by co...@covalent.net.
On Thu, 25 Jul 2002, Peter Donald wrote:
> I can claim that
> ...
> is cleaner, clearer, easier to understand ... yada yada.
>
> It follows our current model where work is described by targets and is
> requires very little effort to migrate into a fully fledged, well modularised
> build file.
>
> Where I don't think targetless build files do the same. Stephane seems to
> agree with me thus my -1 is not invalid.
You -1 is perfectly valid - but it is not a veto, just a negative
vote on a majority decision.
Costin
--
To unsubscribe, e-mail: <ma...@jakarta.apache.org>
For additional commands, e-mail: <ma...@jakarta.apache.org>
Re: Vetoes are void? was Re: [VOTE] target-less build files - counting results
Posted by Peter Donald <pe...@apache.org>.
On Wed, 24 Jul 2002 20:19, Stefan Bodewig wrote:
> On Wed, 24 Jul 2002, Peter Donald <pe...@apache.org> wrote:
> > So far I have have vetoed the change and given reasons which other
> > people have supported.
>
> I've cited all reasons as far as I can tell. These reasons are
> reasons against top-level tasks, not against target-less builds.
I have no problem with top level tasks and I have never said I did as far as I
recall. I would prefer that only declarators were part of top level but as
that is nearly impossible to implement due to ants wonderful codebase I have
no problem with making any task a top level task.
> You can not honestly claim that my example (b) is cleaner, clearer,
> easier to understand or makes the learning curve smoother than my
> example (a).
I can claim that
<project>
<target name="main">
<echo>Hello world</echo>
</target>
</project>
is cleaner, clearer, easier to understand ... yada yada.
It follows our current model where work is described by targets and is
requires very little effort to migrate into a fully fledged, well modularised
build file.
Where I don't think targetless build files do the same. Stephane seems to
agree with me thus my -1 is not invalid.
--
Cheers,
Peter Donald
"Artists can color the sky red because they know it's blue. Those of us who
aren't artists must color things the way they really are or people might
think we're stupid." -- Jules Feiffer
--
To unsubscribe, e-mail: <ma...@jakarta.apache.org>
For additional commands, e-mail: <ma...@jakarta.apache.org>
Re: Vetoes are void? was Re: [VOTE] target-less build files - counting results
Posted by Stefan Bodewig <bo...@apache.org>.
On Wed, 24 Jul 2002, Peter Donald <pe...@apache.org> wrote:
> So far I have have vetoed the change and given reasons which other
> people have supported.
I've cited all reasons as far as I can tell. These reasons are
reasons against top-level tasks, not against target-less builds.
You can not honestly claim that my example (b) is cleaner, clearer,
easier to understand or makes the learning curve smoother than my
example (a).
If you were saying (b) shouldn't be possibe either, that is a
different thing.
> You have failed to convince me to lift the veto. Thus under current
> apache rules the change needs to be reverted. At least thats my
> understanding of http://jakarta.apache.org/site/decisions.html
As it is mine.
But please don't claim that this has always been your understanding
and that you've always been following that rule.
> However you and Costin have decided that no vetos count anymore -
Wrong.
> only majority votes.
I don't think I've said that, and I certainly don't think it is true.
> Because any veto enacts a "revolution" and then that revolution is
> accepted by majority
I said that I cannot find any documents on how revolutions get
accepted. You claimed you knew how (something about accepted by all
but three committers) but failed to say where this knowledge comes
from.
I'm far from declaring anything a revolution, you know that quite
well.
> If so I think this needs to be decided by a far larger community as
> I don't believe your interpretation is common to the rest of
> jakarta.
Before you do, please make sure that you don't claim people would
change the rules when they don't.
Stefan
--
To unsubscribe, e-mail: <ma...@jakarta.apache.org>
For additional commands, e-mail: <ma...@jakarta.apache.org>
Re: Vetoes are void? was Re: [VOTE] target-less build files - counting results
Posted by Stefan Bodewig <bo...@apache.org>.
On Thu, 25 Jul 2002, Peter Donald <pe...@apache.org> wrote:
> I have not voted in the "vote" at all.
<http://marc.theaimsgroup.com/?l=ant-dev&m=102697745627127&w=2>
--
To unsubscribe, e-mail: <ma...@jakarta.apache.org>
For additional commands, e-mail: <ma...@jakarta.apache.org>
Re: Vetoes are void? was Re: [VOTE] target-less build files - counting
results
Posted by co...@covalent.net.
On Thu, 25 Jul 2002, Peter Donald wrote:
> On Thu, 25 Jul 2002 00:33, costinm@covalent.net wrote:
> > On Wed, 24 Jul 2002, Peter Donald wrote:
> > > So far I have have vetoed the change and given reasons which other people
> > > have supported. You have failed to convince me to lift the veto. Thus
> > > under current apache rules the change needs to be reverted. At least
> > > thats my understanding of http://jakarta.apache.org/site/decisions.html
> >
> > This was a [VOTE] - not a commit or code change, but a plan on how
> > the next release should behave.
>
> Actually it was a veto on the code change that was ignored. I have not voted
> in the "vote" at all.
I think this threas is about
[VOTE] target-less build files
And the vote resulted in a majority +1.
You can vote any implementation of this feature - if you have valid
reasons and a better implementation. Any veto on the grounds that this
feature shouldn't be implemented is IMHO invalid.
And if Stefan really want his implementation accepted, I think the
rules for revolutions allow a majority to accept the code in.
But even without 'revolution', I think the requirements for a veto
include a proposal for a better solution to implement the feature
( at least for features that had been voted ).
Costin
--
To unsubscribe, e-mail: <ma...@jakarta.apache.org>
For additional commands, e-mail: <ma...@jakarta.apache.org>
Re: Vetoes are void? was Re: [VOTE] target-less build files - counting results
Posted by Peter Donald <pe...@apache.org>.
On Thu, 25 Jul 2002 00:33, costinm@covalent.net wrote:
> On Wed, 24 Jul 2002, Peter Donald wrote:
> > So far I have have vetoed the change and given reasons which other people
> > have supported. You have failed to convince me to lift the veto. Thus
> > under current apache rules the change needs to be reverted. At least
> > thats my understanding of http://jakarta.apache.org/site/decisions.html
>
> This was a [VOTE] - not a commit or code change, but a plan on how
> the next release should behave.
Actually it was a veto on the code change that was ignored. I have not voted
in the "vote" at all.
--
Cheers,
Peter Donald
*------------------------------------------------------*
| Despite your efforts to be a romantic hero, you will |
| gradually evolve into a postmodern plot device. |
*------------------------------------------------------*
--
To unsubscribe, e-mail: <ma...@jakarta.apache.org>
For additional commands, e-mail: <ma...@jakarta.apache.org>
Re: Vetoes are void? was Re: [VOTE] target-less build files - counting
results
Posted by Sam Ruby <ru...@apache.org>.
costinm@covalent.net wrote:
>
> I strongly believe that it's the majority of commiters who should
> decide what and how is released
[snip]
> I think a veto on a plan doesn't even exist - it's not a veto, but
> a negative vote in a majority voting.
+1 on both counts.
http://jakarta.apache.org/site/decisions.html
http://jakarta.apache.org/site/pmc/01-01-17-meeting-minutes.html
- Sam Ruby
--
To unsubscribe, e-mail: <ma...@jakarta.apache.org>
For additional commands, e-mail: <ma...@jakarta.apache.org>
Re: Vetoes are void? was Re: [VOTE] target-less build files - counting
results
Posted by co...@covalent.net.
On Wed, 24 Jul 2002, Peter Donald wrote:
> So far I have have vetoed the change and given reasons which other people have
> supported. You have failed to convince me to lift the veto. Thus under
> current apache rules the change needs to be reverted. At least thats my
> understanding of http://jakarta.apache.org/site/decisions.html
This was a [VOTE] - not a commit or code change, but a plan on how
the next release should behave.
I strongly believe that it's the majority of commiters who should
decide what and how is released - not a single individual to control
it ( by blocking everything ).
> However you and Costin have decided that no vetos count anymore - only
> majority votes. Because any veto enacts a "revolution" and then that
> revolution is accepted by majority - thats the reasoning I believe?
Fact is that the majority of commiters:
- can release anything it wants. ( release rules )
- can use any codebase it wants. ( revolution rules ).
That's pretty clear.
Vetoes with reason and a proposed soultion are perfectly valid
for code changes - but the current rules are pretty clear IMHO that
the majority of votes can override this via revolutions.
If someone makes a commit - it can be vetoed.
If a new features is decided by the majority - you can veto any
implementation ( as long as you provide a better one or enough
reasons on why is bad )
> If so I think this needs to be decided by a far larger community as I don't
> believe your interpretation is common to the rest of jakarta. If this is your
> interpretation then I will take it to the general/pmc lists and discuss it
> there and see if the jakarta rules can be changed.
I don't think any rule have to be changed - but made clearer.
I think the veto is greatly abused - you veto changes in the main branch
while your revolution does similar things.
> However what I do want for is this be consistent. If one veto is ruled to be
> invlaid then all vetoes become invalid become invalid.
I think a veto on a plan doesn't even exist - it's not a veto, but
a negative vote in a majority voting.
Costin
--
To unsubscribe, e-mail: <ma...@jakarta.apache.org>
For additional commands, e-mail: <ma...@jakarta.apache.org>
Vetoes are void? was Re: [VOTE] target-less build files - counting results
Posted by Peter Donald <pe...@apache.org>.
On Wed, 24 Jul 2002 00:35, Stefan Bodewig wrote:
> On Tue, 23 Jul 2002, <co...@covalent.net> wrote:
> > It doesn't matter.
>
> It may not matter to you.
>
> > I consider the subject closed,
>
> So do I, but we do so for different reasons.
>
> > ant will support targetless build files.
>
> Yes.
So far I have have vetoed the change and given reasons which other people have
supported. You have failed to convince me to lift the veto. Thus under
current apache rules the change needs to be reverted. At least thats my
understanding of http://jakarta.apache.org/site/decisions.html
However you and Costin have decided that no vetos count anymore - only
majority votes. Because any veto enacts a "revolution" and then that
revolution is accepted by majority - thats the reasoning I believe?
If so I think this needs to be decided by a far larger community as I don't
believe your interpretation is common to the rest of jakarta. If this is your
interpretation then I will take it to the general/pmc lists and discuss it
there and see if the jakarta rules can be changed.
However what I do want for is this be consistent. If one veto is ruled to be
invlaid then all vetoes become invalid become invalid.
--
Cheers,
Peter Donald
*------------------------------------------------*
| The student who is never required to do what |
| he cannot do never does what he can do. |
| - John Stuart Mill |
*------------------------------------------------*
--
To unsubscribe, e-mail: <ma...@jakarta.apache.org>
For additional commands, e-mail: <ma...@jakarta.apache.org>
Re: [VOTE] target-less build files - counting results
Posted by Stefan Bodewig <bo...@apache.org>.
On Tue, 23 Jul 2002, <co...@covalent.net> wrote:
> It doesn't matter.
It may not matter to you.
> I consider the subject closed,
So do I, but we do so for different reasons.
> ant will support targetless build files.
Yes.
Stefan
--
To unsubscribe, e-mail: <ma...@jakarta.apache.org>
For additional commands, e-mail: <ma...@jakarta.apache.org>
Re: [VOTE] target-less build files - counting results
Posted by Nicola Ken Barozzi <ni...@apache.org>.
Peter Donald wrote:
> On Wed, 24 Jul 2002 00:26, costinm@covalent.net wrote:
>
>>Peter can propose another vote if he wants - but until this happens
>>I consider the subject closed, ant will support targetless build
>>files.
>
> Actually Peter already did just that. It was vetoed ... much the same way that
> target-less build files was vetoed. Stefan reverted the change. It seems the
> rules change depending on who is doing the veto.
Peter, I agree with you about targetless buildfiles...
But I don't see why it really has to be vetoed.
If users want to shoot themselves in the foot, let them do it.
Targetless buildfile support will not prevent us from using only
targets, as in Java having static code snippets doesn't prevent one from
writing correct methods.
--
Nicola Ken Barozzi nicolaken@apache.org
- verba volant, scripta manent -
(discussions get forgotten, just code remains)
---------------------------------------------------------------------
--
To unsubscribe, e-mail: <ma...@jakarta.apache.org>
For additional commands, e-mail: <ma...@jakarta.apache.org>
Re: [VOTE] target-less build files - counting results
Posted by Stefan Bodewig <bo...@apache.org>.
On Wed, 24 Jul 2002, Peter Donald <pe...@apache.org> wrote:
>> There has been no vote,
>
> there was originally.
I can't find it.
Stefan
--
To unsubscribe, e-mail: <ma...@jakarta.apache.org>
For additional commands, e-mail: <ma...@jakarta.apache.org>
Re: [VOTE] target-less build files - counting results
Posted by Peter Donald <pe...@apache.org>.
On Wed, 24 Jul 2002 19:58, Stefan Bodewig wrote:
> On Wed, 24 Jul 2002, Peter Donald <pe...@apache.org> wrote:
> > On Wed, 24 Jul 2002 00:26, costinm@covalent.net wrote:
> >> Peter can propose another vote if he wants - but until this happens
> >> I consider the subject closed, ant will support targetless build
> >> files.
> >
> > Actually Peter already did just that. It was vetoed ... much the
> > same way that target-less build files was vetoed. Stefan reverted
> > the change.
>
> Huh? You are talking about making <project>'s default attribute
> default to "main" if omitted, yes?
yep.
> There has been no vote,
there was originally.
> It seems the situation is a bit different.
hmmmm ... yes, now it is Stefans change.
> > It seems the rules change depending on who is doing the veto.
>
> It seems the rules change depending on who is interpreting them, yes,
> this it too true.
>
> I know you've been advocating over and over again that a -1 becomes
> invalid if all reasons have been addressed. I am playing by your
> rules.
You have not addressed my reasons at all.
--
Cheers,
Peter Donald
--------------------------------
My opinions may have changed,
but not the fact that I am right
--------------------------------
--
To unsubscribe, e-mail: <ma...@jakarta.apache.org>
For additional commands, e-mail: <ma...@jakarta.apache.org>
Re: [VOTE] target-less build files - counting results
Posted by Stefan Bodewig <bo...@apache.org>.
On Wed, 24 Jul 2002, Peter Donald <pe...@apache.org> wrote:
> On Wed, 24 Jul 2002 00:26, costinm@covalent.net wrote:
>> Peter can propose another vote if he wants - but until this happens
>> I consider the subject closed, ant will support targetless build
>> files.
>
> Actually Peter already did just that. It was vetoed ... much the
> same way that target-less build files was vetoed. Stefan reverted
> the change.
Huh? You are talking about making <project>'s default attribute
default to "main" if omitted, yes?
There has been no vote, Conor gave a -1 with a reason, the reason has
been confirmed by Steve and you never responded.
It seems the situation is a bit different.
> It seems the rules change depending on who is doing the veto.
It seems the rules change depending on who is interpreting them, yes,
this it too true.
I know you've been advocating over and over again that a -1 becomes
invalid if all reasons have been addressed. I am playing by your
rules.
Stefan
--
To unsubscribe, e-mail: <ma...@jakarta.apache.org>
For additional commands, e-mail: <ma...@jakarta.apache.org>
Re: [VOTE] target-less build files - counting results
Posted by Peter Donald <pe...@apache.org>.
On Wed, 24 Jul 2002 00:26, costinm@covalent.net wrote:
> Peter can propose another vote if he wants - but until this happens
> I consider the subject closed, ant will support targetless build
> files.
Actually Peter already did just that. It was vetoed ... much the same way that
target-less build files was vetoed. Stefan reverted the change. It seems the
rules change depending on who is doing the veto.
--
Cheers,
Peter Donald
----------------------------------------------------------------
Fools ignore complexity. Pragmatists suffer it.
Some can avoid it. Geniuses remove it.
-- Perlis's Programming Proverb #58, SIGPLAN Notices, Sept. 1982
----------------------------------------------------------------
--
To unsubscribe, e-mail: <ma...@jakarta.apache.org>
For additional commands, e-mail: <ma...@jakarta.apache.org>
Re: [VOTE] target-less build files - counting results
Posted by co...@covalent.net.
On 23 Jul 2002, Stefan Bodewig wrote:
> We've had six +1s and one -1 by the committers who did not abstain
> (and Stephane's explicit 0).
> Comparing (a) to (b) the first argument is not valid IMHO, I cannot
It doesn't matter.
The decision to remove top-level tasks had few -1s ( mine included ),
but it was a majority vote.
The current rules for release and 'rule for revolutionary' imply that
the majority can decide what features and code it wants ( instead
of one individual making the decisions )
Peter can propose another vote if he wants - but until this happens
I consider the subject closed, ant will support targetless build
files.
Costin
> see how (b) is clearer than (a). Also I fail to see how the evolution
> from (b) to good and modular build files may be easier than from (a).
>
> Peter, any other reasons?
>
> Stefan
>
> Footnotes:
> [1] <http://marc.theaimsgroup.com/?l=ant-dev&m=102697745627127&w=2>
>
>
> --
> To unsubscribe, e-mail: <ma...@jakarta.apache.org>
> For additional commands, e-mail: <ma...@jakarta.apache.org>
>
>
--
To unsubscribe, e-mail: <ma...@jakarta.apache.org>
For additional commands, e-mail: <ma...@jakarta.apache.org>
[VOTE] target-less build files - counting results
Posted by Stefan Bodewig <bo...@apache.org>.
On Sun, 21 Jul 2002, <co...@covalent.net> wrote:
> I'm getting tired of this discussion. I don't know if someone is
> counting the votes
I've done so.
We've had six +1s and one -1 by the committers who did not abstain
(and Stephane's explicit 0).
Most arguments we've seen are arguments against task at the top level
in general, not arguments against target less build files, which means
they are irrelevant for this vote. ant-dev has agreed by lazy
consensus that we want to allow all tasks at the top level, if anybody
wants to make the vote explicit, feel free to do so.
Having said that, we are faced with the choice between
(a)
<project>
<echo>Hello world</echo>
</project>
and
(b)
<project default="foo">
<echo>Hello world</echo>
<target name="foo" />
</project>
(b) is possible no matter how this vote ends.
Peter's reasons for his -1 are [1]:
,----
| Removes clarity and structure from build file. The evolution from
| target-less to targetful build process can also be ugly.
`----
Comparing (a) to (b) the first argument is not valid IMHO, I cannot
see how (b) is clearer than (a). Also I fail to see how the evolution
from (b) to good and modular build files may be easier than from (a).
Peter, any other reasons?
Stefan
Footnotes:
[1] <http://marc.theaimsgroup.com/?l=ant-dev&m=102697745627127&w=2>
--
To unsubscribe, e-mail: <ma...@jakarta.apache.org>
For additional commands, e-mail: <ma...@jakarta.apache.org>
Re: [VOTE] target-less build files (plus a new proposal)
Posted by Nicola Ken Barozzi <ni...@apache.org>.
costinm@covalent.net wrote:
> On Sun, 19 May 2002, Ara Abrahamian wrote:
>
>
>>It's like a book without chapters :-) Boring, non-modular.
>
> Maybe it's a short article - direct and on the subject.
>
> You don't need 100 page book with chapters, appendix, eratta
> to compile 2 java files.
Nonsense
<project>
<init>
<doit/>
<doitagain/>
</init>
</project>
<project>
<doit/>
<doitagain/>
</project>
Tell me why the first is a 100 page book and the second a short article.
Nonsense.
We need to have stuff run before every build and after every build.
If you really are bothered by the latter, use the former, but let's get
over with it.
>>I really don't find a target-less build file useful. Ant's build files
>
> What I fail to understand is why people who don't find something useful
> for themself assume that nobody will need them and can be removed.
Look, it's not a matter of features, but a matter of semantics.
> Or why people assume what they like is 'good practice' and what
> other people like is 'bad practice'.
It's just more clear.
> Nicola spends more time trying to deprecate top-level targets then he
> spends trying to get <init>.
Because with top-level tasks <init> is not needed.
They are the *same* thing.
I'm just saying that putting all top-level stuff inside an <init> tag is
cleaner, etc, blah blah blah.
But I honestly don't care, if you really *hate* having to put them
inside another tag put them top level, no problem.
Just see if you can get the -1s away; I'm not a committer, and I would
vote -0.
Other committers are -1, and I tried to get to a compromise (init) that
*could* have made things work for all.
If you are so stiff, go ahead, but you're not getting collaboration from
other committers.
Good luck.
--
Nicola Ken Barozzi nicolaken@apache.org
- verba volant, scripta manent -
(discussions get forgotten, just code remains)
---------------------------------------------------------------------
--
To unsubscribe, e-mail: <ma...@jakarta.apache.org>
For additional commands, e-mail: <ma...@jakarta.apache.org>
RE: [VOTE] target-less build files (plus a new proposal)
Posted by co...@covalent.net.
On Mon, 20 May 2002, Ara Abrahamian wrote:
> Hey! It's what I think. I think it's bad. You think it's good. Maybe I'm
> wrong, if many ppl think it's good. It's just a technical debate.
It doesn't look like a technical debate - we are arguing over taste
and style. There is little technical here.
Some people are saying 'we like that and want to use it'. Other people are
saying "we don't like that and we'll not use it".
The problem is they are also saying - "because we don't like it, you
shouldn't use it as well".
I don't think ant's goal is to be perfect ( acording to anyone's taste ),
but to be flexible and open enough to meet everyone's needs.
Costin
--
To unsubscribe, e-mail: <ma...@jakarta.apache.org>
For additional commands, e-mail: <ma...@jakarta.apache.org>
RE: [VOTE] target-less build files (plus a new proposal)
Posted by Ara Abrahamian <ar...@yahoo.com>.
> Well, do you have a magic ball that tells you what's good
> practice ?
>
> It seems there are at least few ant commiters who seem to believe
> that targetless build files are usefull and convenient, and
> a valid ( and good ) practice. I don't think we're stupid, and
> the fact that we used ant for quite a bit of time and several
> projects sugests we may not be completely clueless.
>
> I'm getting tired of this discussion. I don't know if someone is
> counting the votes - probably we need another [VOTE] thread
> with the 2 separate issues, top level tasks and targetless
> builds. And the majority should decide - no need to repeat
> arguments over and over.
Hey! It's what I think. I think it's bad. You think it's good. Maybe I'm
wrong, if many ppl think it's good. It's just a technical debate.
Ara.
--
To unsubscribe, e-mail: <ma...@jakarta.apache.org>
For additional commands, e-mail: <ma...@jakarta.apache.org>
RE: [VOTE] target-less build files (plus a new proposal)
Posted by co...@covalent.net.
On Sun, 19 May 2002, Ara Abrahamian wrote:
> It's like a book without chapters :-) Boring, non-modular.
Maybe it's a short article - direct and on the subject.
You don't need 100 page book with chapters, appendix, eratta
to compile 2 java files.
> I really don't find a target-less build file useful. Ant's build files
What I fail to understand is why people who don't find something useful
for themself assume that nobody will need them and can be removed.
Or why people assume what they like is 'good practice' and what
other people like is 'bad practice'.
Nicola spends more time trying to deprecate top-level targets then he
spends trying to get <init>.
> names like "main" are generally used everywhere as the base target. And
> imho, like Java language itself, we should encourage ppl to use good
> practices and even disallow bad practices, or leave them out of the
> system at least, like Java.
Well, do you have a magic ball that tells you what's good
practice ?
It seems there are at least few ant commiters who seem to believe
that targetless build files are usefull and convenient, and
a valid ( and good ) practice. I don't think we're stupid, and
the fact that we used ant for quite a bit of time and several
projects sugests we may not be completely clueless.
I'm getting tired of this discussion. I don't know if someone is
counting the votes - probably we need another [VOTE] thread
with the 2 separate issues, top level tasks and targetless
builds. And the majority should decide - no need to repeat
arguments over and over.
Costin
--
To unsubscribe, e-mail: <ma...@jakarta.apache.org>
For additional commands, e-mail: <ma...@jakarta.apache.org>
RE: [VOTE] target-less build files (plus a new proposal)
Posted by Diane Holt <ho...@yahoo.com>.
--- Ara Abrahamian <ar...@yahoo.com> wrote:
> It's like a book without chapters :-)
Actually, it's more like: Now you can write a short story, if that's all
the story you have to tell; or a novella, if you have a bit more; or a
novel, if you've got even more; or even a "A la recherche du temps perdu",
if you've got lots and lots of stories to tell :)
> With targets you have a simple mechanism to define small functions in
> your build, call them, depend on them.
If you have a need to, sure. But why should we require that?
> Like you "extract variable" in refactoring of your java code, you
> also "extract target" in your build file, you give it a good name
> and the build file is more readable then.
I don't happen to see it that way. I find a sequential list of tasks to
perform just as (and in some cases, more) readable.
> I really don't find a target-less build file useful.
Then you likely wouldn't write one.
> [proposal for reusable build files]
> So what do you think?
I think you should probably start it in a separate thread so it doesn't
get buried in this one.
Diane
=====
(holtdl@yahoo.com)
__________________________________________________
Do You Yahoo!?
Yahoo! Health - Feel better, live better
http://health.yahoo.com
--
To unsubscribe, e-mail: <ma...@jakarta.apache.org>
For additional commands, e-mail: <ma...@jakarta.apache.org>