NetBeans, Scala, jars

On Thu, Apr 16, 2009 at 02:25:58PM -0700, Charles F. Munat wrote:
> I'm trying to use Berkeley DB XML with a Scala GUI project. I've
> imported the jars into Netbeans (same as I did with JFreeChart, which
> I'm also using), but when I try to import from them, I get the red
> exclamation mark and an error message that says, "value sleepycat is
> not a member of package com."

Every time someone asks about this, god kills a kitten!

It's not your fault charles, it just makes me sad to think this trap is
going to be snapping down on peoples' legs for the rest of our lives,
and all we can really do is give you a hacksaw so you can saw it off a
little quicker.

import _root_.com.sleepycat.what.ev.er

A little more background on it.
Scala's imports are relative, and I bet 100 Ukrainian dollars that you have package au.com.munat or something at the top of this source file.  So import com.sleepycat.Foo actually imports au.com.sleepycat.Foo, which doesn't exist.
Martin etc. have apparently decided that no amount of protestation is going to change this.  Perhaps Miles can make Eclipse automatically prepend the _root_ or hint that it should be prepended.
2009/4/16 Paul Phillips <paulp@improving.org>

On Thu, Apr 16, 2009 at 02:25:58PM -0700, Charles F. Munat wrote:
> I'm trying to use Berkeley DB XML with a Scala GUI project. I've
> imported the jars into Netbeans (same as I did with JFreeChart, which
> I'm also using), but when I try to import from them, I get the red
> exclamation mark and an error message that says, "value sleepycat is
> not a member of package com."

Every time someone asks about this, god kills a kitten!

It's not your fault charles, it just makes me sad to think this trap is
going to be snapping down on peoples' legs for the rest of our lives,
and all we can really do is give you a hacksaw so you can saw it off a
little quicker.

 import _root_.com.sleepycat.what.ev.er

--
Paul Phillips      | It's better to have gloved and tossed than never to
Caged Spirit       | have played baseball.
Empiricist         |
ha! spill, pupil   |----------* http://www.improving.org/paulp/ *----------

I'll be pillared, but I've come around to the truly like relative import.  I'd actually hate to see them go.  Would be nicer if the absolute root identifier was more succinct and less offensive to one's ascetics. 

ie.
import .com.sleepycat.Foo
import _|_.com.sleepycat.Foo
import !.com.sleepycat.Foo

But I'll take warty_root_ and to keep relative paths. 

On Thu, Apr 16, 2009 at 6:17 PM, Ricky Clarkson <ricky.clarkson@gmail.com> wrote:

A little more background on it.
Scala's imports are relative, and I bet 100 Ukrainian dollars that you have package au.com.munat or something at the top of this source file.  So import com.sleepycat.Foo actually imports au.com.sleepycat.Foo, which doesn't exist.
Martin etc. have apparently decided that no amount of protestation is going to change this.  Perhaps Miles can make Eclipse automatically prepend the _root_ or hint that it should be prepended.
2009/4/16 Paul Phillips <paulp@improving.org>

On Thu, Apr 16, 2009 at 02:25:58PM -0700, Charles F. Munat wrote:
> I'm trying to use Berkeley DB XML with a Scala GUI project. I've
> imported the jars into Netbeans (same as I did with JFreeChart, which
> I'm also using), but when I try to import from them, I get the red
> exclamation mark and an error message that says, "value sleepycat is
> not a member of package com."

Every time someone asks about this, god kills a kitten!

It's not your fault charles, it just makes me sad to think this trap is
going to be snapping down on peoples' legs for the rest of our lives,
and all we can really do is give you a hacksaw so you can saw it off a
little quicker.

 import _root_.com.sleepycat.what.ev.er

--
Paul Phillips      | It's better to have gloved and tossed than never to
Caged Spirit       | have played baseball.
Empiricist         |
ha! spill, pupil   |----------* http://www.improving.org/paulp/ *----------

On Thursday April 16 2009, Ray Racine wrote:
> I'll be pillared, but I've come around to the truly like relative
> import. I'd actually hate to see them go. Would be nicer if the
> absolute root identifier was more succinct and less offensive to
> one's ascetics.

Please, leave my ascetics out of it. They're busy making themselves
miserable.

> ie.
> import .com.sleepycat.Foo
> import _|_.com.sleepycat.Foo
> import !.com.sleepycat.Foo
>
> But I'll take warty_root_ and to keep relative paths.

As to aesthetic concerns, I, too, find _root_ unpleasant. I might go so
far as to advocate using a simple ".", analogously to file system path
names in Unix-style file systems.

Randall Schulz

Randall R Schulz wrote:
> As to aesthetic concerns, I, too, find _root_ unpleasant. I might go so
> far as to advocate using a simple ".", analogously to file system path
> names in Unix-style file systems.

+1. Another precedent is the :: prefix in C++ to access the global
namespace.

I taught Java to novices for a couple of years.  Most common beginning mistake in writing Swing programs?
import.javax.swing.JFrame;  In case it's not obvious, I'll add some whitespace: import .javax.swing.JFrame;

I wouldn't want to make this accident actually valid code. :)
I should make a script that automatically replies to messages mentioning Eclipse, Netbeans or IDEA, saying "does the problem persist without the IDE?". :)
2009/4/17 Randall R Schulz <rschulz@sonic.net>

On Thursday April 16 2009, Ray Racine wrote:
> I'll be pillared, but I've come around to the truly like relative
> import. I'd actually hate to see them go.  Would be nicer if the
> absolute root identifier was more succinct and less offensive to
> one's ascetics.

Please, leave my ascetics out of it. They're busy making themselves
miserable.


> ie.
> import .com.sleepycat.Foo
> import _|_.com.sleepycat.Foo
> import !.com.sleepycat.Foo
>
> But I'll take warty_root_ and to keep relative paths.


As to aesthetic concerns, I, too, find _root_ unpleasant. I might go so
far as to advocate using a simple ".", analogously to file system path
names in Unix-style file systems.


Randall Schulz

Also in domain names the root is "." so you can write "google.com." (note the trailing dot). And in the JVM world packages often follow domain names (but reversed). So the leading "." would make sense in that sense too.

-Arthur


On Thu, Apr 16, 2009 at 9:49 PM, Eric Willigers <ewilligers@gmail.com> wrote:

Randall R Schulz wrote:

As to aesthetic concerns, I, too, find _root_ unpleasant. I might go so far as to advocate using a simple ".", analogously to file system path  names in Unix-style file systems.

+1. Another precedent is the :: prefix in C++ to access the global namespace.

> From: Ricky Clarkson [mailto:ricky.clarkson@gmail.com]
> I taught Java to novices for a couple of years. Most common
> beginning mistake in writing Swing programs?
>
> import.javax.swing.JFrame; In case it's not obvious, I'll
> add some whitespace:
> import .javax.swing.JFrame;
>
>
> I wouldn't want to make this accident actually valid code. :)

Why not?
In Java it is a mistake simply because it's wrong.

In any other language, if it is valid code, it is
simply no mistake.

> I should make a script that automatically replies to messages
> mentioning Eclipse, Netbeans or IDEA, saying "does the
> problem persist without the IDE?". :)

:-) +1 here
(BTW: I wanted something like that in a Java list,
where every improvent to Java The Language was answered
with "Why? The IDE makes it sooo easy right now").

I think the tendency of the people here is indeed to pick leading `.' for root.

Cheers

On Fri, Apr 17, 2009 at 2:55 AM, Ricky Clarkson
wrote:
> I should make a script that automatically replies to messages mentioning
> Eclipse, Netbeans or IDEA, saying "does the problem persist without the
> IDE?". :)

Yes! Please!

Cheers,

Miles

Quoting Ricky Clarkson :

> I taught Java to novices for a couple of years. Most common beginning
> mistake in writing Swing programs?
> import.javax.swing.JFrame; In case it's not obvious, I'll add some
> whitespace:
> import .javax.swing.JFrame;
>
> I wouldn't want to make this accident actually valid code. :)

Uh, why wouldn't you want to make this "accident" valid code?
If the root package's package javax 's package swing 's class JFrame
is wanted, and root could be written as ".", then your example would
actually be valid, and no mistake, and do exactly what was intended.
So why's that bad?

-Martin

* Martin S. Weber [2009-04-17 09:54 -0400]:
> Quoting Ricky Clarkson :
>> import.javax.swing.JFrame;
>>
>> I wouldn't want to make this accident actually valid code. :)
>
> Uh, why wouldn't you want to make this "accident" valid code?
> If the root package's package javax 's package swing 's class JFrame is
> wanted, and root could be written as ".", then your example would
> actually be valid, and no mistake, and do exactly what was intended. So
> why's that bad?

I would further add that if someone typed this erroneously (i.e. they
intended to use a relative package name), it would simply be the inverse
of the current erroneous situation (typing an absolute package when a
relative one is needed) and the compiler would catch it.

I wouldn't want accidents to be valid code.  I'm surprised this is controversial.   2009/4/17 Phil! Gold <phil_g@pobox.com>

* Martin S. Weber <martin.weber@nist.gov> [2009-04-17 09:54 -0400]:
> Quoting Ricky Clarkson <ricky.clarkson@gmail.com>:
>> import.javax.swing.JFrame;
>>
>> I wouldn't want to make this accident actually valid code. :)
>
> Uh, why wouldn't you want to make this "accident" valid code?
> If the root package's package javax 's package swing 's class JFrame is
> wanted, and root could be written as ".", then your example would
> actually be valid, and no mistake, and do exactly what was intended. So
> why's that bad?

I would further add that if someone typed this erroneously (i.e. they
intended to use a relative package name), it would simply be the inverse
of the current erroneous situation (typing an absolute package when a
relative one is needed) and the compiler would catch it.

--
...computer contrarian of the first order... / http://aperiodic.net/phil/
PGP: 026A27F2  print: D200 5BDB FC4B B24A 9248  9F7A 4322 2D22 026A 27F2

Quoting Ricky Clarkson :

> I wouldn't want accidents to be valid code. I'm surprised this is
> controversial.

I think the intent of my message did not make it over clear. This is
no "accident" (<-- quotes!). It's a datapoint of people intuitively
referring to the root package with a leading dot :-)

A lack of semicolon to end the line is an "accident" in java, too,
while it is perfectly valid in scala. Is this ability to be banned in
your eyes, too?

-Martin

+1

On 4/16/09, Eric Willigers wrote:
> Randall R Schulz wrote:
>
> > As to aesthetic concerns, I, too, find _root_ unpleasant. I might go so
> far as to advocate using a simple ".", analogously to file system path
> names in Unix-style file systems.
> >
>
> +1. Another precedent is the :: prefix in C++ to access the global
> namespace.
>
>
>

On Fri, Apr 17, 2009 at 2:54 PM, Ricky Clarkson <ricky.clarkson@gmail.com> wrote:

I wouldn't want accidents to be valid code.  I'm surprised this is controversial.

Look at it this way: one goal of a good language is generally DWIM ("do what I mean"): one's intuitions of "this oughtta work" should be correct as often as possible.  (Obviously, that has to be balanced against other concerns, but it's a goal.) 

So these sorts of common mistakes are actually valuable information about what people consider intuitive.  Indeed, it's quite encouraging that your naive students want to do the same thing that the consensus of more experienced engineers here seems to be leaning towards...

Except that they're not doing it on purpose, and if Java compiled import.javax.swing.JFrame to mean something different to import javax.swing.JFrame, they would use the wrong one sometimes and find it hard to remember whether to use dotless or dotty syntax for each case.  Novices don't notice punctuation as much as we might, and we can all still misread sometimes.  Java (and current Scala) do the right thing, they give an error message.

2009/4/18 Justin du coeur <jducoeur@gmail.com>

On Fri, Apr 17, 2009 at 2:54 PM, Ricky Clarkson <ricky.clarkson@gmail.com> wrote:

I wouldn't want accidents to be valid code.  I'm surprised this is controversial.

Look at it this way: one goal of a good language is generally DWIM ("do what I mean"): one's intuitions of "this oughtta work" should be correct as often as possible.  (Obviously, that has to be balanced against other concerns, but it's a goal.) 

So these sorts of common mistakes are actually valuable information about what people consider intuitive.  Indeed, it's quite encouraging that your naive students want to do the same thing that the consensus of more experienced engineers here seems to be leaning towards...

I have a hard time constructing a real life example where refusing the
dot-prefixed syntax would help detecting any bug.

Even then, it would probably be trivial to emit a warning if the absolute
syntax shadows a relative path, the probability of which is close to
zero, IMO.

On 4/20/09, Ricky Clarkson wrote:
> Except that they're not doing it on purpose, and if Java compiled
> import.javax.swing.JFrame to mean something different to import
> javax.swing.JFrame, they would use the wrong one sometimes and find it hard
> to remember whether to use dotless or dotty syntax for each case. Novices
> don't notice punctuation as much as we might, and we can all still misread
> sometimes. Java (and current Scala) do the right thing, they give an error
> message.
>
> 2009/4/18 Justin du coeur
>
> >
> >
> > On Fri, Apr 17, 2009 at 2:54 PM, Ricky Clarkson
> wrote:
> >
> > >
> > > I wouldn't want accidents to be valid code. I'm surprised this is
> controversial.
> >
> > Look at it this way: one goal of a good language is generally DWIM ("do
> what I mean"): one's intuitions of "this oughtta work" should be correct as
> often as possible. (Obviously, that has to be balanced against other
> concerns, but it's a goal.)
> >
> > So these sorts of common mistakes are actually valuable information about
> what people consider intuitive. Indeed, it's quite encouraging that your
> naive students want to do the same thing that the consensus of more
> experienced engineers here seems to be leaning towards...
> >
>
>

If the probability of such a shadowing is close to zero, then how about just making the existing import syntax able to drop down to absolute import when the relative import cannot be resolved?

package uk.org.ourcorp

import org.apache.Stuff

There is no uk.org.apache.Stuff, so let's backtrack and drop the uk bit.

If the probability of such a shadowing is not sufficiently close to zero for the above, then perhaps I have a point. :)

2009/4/21 Christian Szegedy <christian.szegedy@gmail.com>

I have a hard time constructing a real life example where refusing the
dot-prefixed syntax would help detecting any bug.

Even then, it would probably be trivial to emit a warning if the absolute
syntax shadows a relative path, the probability of which is close to
zero, IMO.

On 4/20/09, Ricky Clarkson <ricky.clarkson@gmail.com> wrote:
> Except that they're not doing it on purpose, and if Java compiled
> import.javax.swing.JFrame to mean something different to import
> javax.swing.JFrame, they would use the wrong one sometimes and find it hard
> to remember whether to use dotless or dotty syntax for each case.  Novices
> don't notice punctuation as much as we might, and we can all still misread
> sometimes.  Java (and current Scala) do the right thing, they give an error
> message.
>
> 2009/4/18 Justin du coeur <jducoeur@gmail.com>
>
> >
> >
> > On Fri, Apr 17, 2009 at 2:54 PM, Ricky Clarkson <ricky.clarkson@gmail.com>
> wrote:
> >
> > >
> > > I wouldn't want accidents to be valid code.  I'm surprised this is
> controversial.
> >
> > Look at it this way: one goal of a good language is generally DWIM ("do
> what I mean"): one's intuitions of "this oughtta work" should be correct as
> often as possible.  (Obviously, that has to be balanced against other
> concerns, but it's a goal.)
> >
> > So these sorts of common mistakes are actually valuable information about
> what people consider intuitive.  Indeed, it's quite encouraging that your
> naive students want to do the same thing that the consensus of more
> experienced engineers here seems to be leaning towards...
> >
>
>

View the leading dot as a compiler checked confirmation for the
really thorough reader: "Yes, I really want an absolute path".

On 4/21/09, Ricky Clarkson wrote:
> If the probability of such a shadowing is close to zero, then how about just
> making the existing import syntax able to drop down to absolute import when
> the relative import cannot be resolved?
>
> package uk.org.ourcorp
>
> import org.apache.Stuff
>
> There is no uk.org.apache.Stuff, so let's backtrack and drop the uk bit.
>
> If the probability of such a shadowing is not sufficiently close to zero for
> the above, then perhaps I have a point. :)
>
> 2009/4/21 Christian Szegedy
>
> > I have a hard time constructing a real life example where refusing the
> > dot-prefixed syntax would help detecting any bug.
> >
> > Even then, it would probably be trivial to emit a warning if the absolute
> > syntax shadows a relative path, the probability of which is close to
> > zero, IMO.
> >
> >
> >
> >
> > On 4/20/09, Ricky Clarkson wrote:
> > > Except that they're not doing it on purpose, and if Java compiled
> > > import.javax.swing.JFrame to mean something different to import
> > > javax.swing.JFrame, they would use the wrong one sometimes and find it
> hard
> > > to remember whether to use dotless or dotty syntax for each case.
> Novices
> > > don't notice punctuation as much as we might, and we can all still
> misread
> > > sometimes. Java (and current Scala) do the right thing, they give an
> error
> > > message.
> > >
> > > 2009/4/18 Justin du coeur
> > >
> > > >
> > > >
> > > > On Fri, Apr 17, 2009 at 2:54 PM, Ricky Clarkson
>
> > > wrote:
> > > >
> > > > >
> > > > > I wouldn't want accidents to be valid code. I'm surprised this is
> > > controversial.
> > > >
> > > > Look at it this way: one goal of a good language is generally DWIM
> ("do
> > > what I mean"): one's intuitions of "this oughtta work" should be correct
> as
> > > often as possible. (Obviously, that has to be balanced against other
> > > concerns, but it's a goal.)
> > > >
> > > > So these sorts of common mistakes are actually valuable information
> about
> > > what people consider intuitive. Indeed, it's quite encouraging that
> your
> > > naive students want to do the same thing that the consensus of more
> > > experienced engineers here seems to be leaning towards...
> > > >
> > >
> > >
> >
>
>

I've stated this before, but requiring a marker for a root-relative
import is something that every Java developer who ever comes to the
language will trip over. Probably multiple times. Relative imports are
great, but it is *they* which should be prefixed.

On Fri, Apr 17, 2009 at 1:47 AM, martin odersky wrote:
> I think the tendency of the people here is indeed to pick leading `.' for root.
>
> Cheers
>
>  -- Martin
>
> On Fri, Apr 17, 2009 at 8:48 AM, Detering Dirk
> wrote:
>>> From: Ricky Clarkson [mailto:ricky.clarkson@gmail.com]
>>> I taught Java to novices for a couple of years.  Most common
>>> beginning mistake in writing Swing programs?
>>>
>>> import.javax.swing.JFrame;  In case it's not obvious, I'll
>>> add some whitespace:
>>> import .javax.swing.JFrame;
>>>
>>>
>>> I wouldn't want to make this accident actually valid code. :)
>>
>> Why not?
>> In Java it is a mistake simply because it's wrong.
>>
>> In any other language, if it is valid code, it is
>> simply no mistake.
>>
>>> I should make a script that automatically replies to messages
>>> mentioning Eclipse, Netbeans or IDEA, saying "does the
>>> problem persist without the IDE?". :)
>>
>> :-)  +1 here
>> (BTW: I wanted something like that in a Java list,
>> where every improvent to Java The Language was answered
>> with "Why? The IDE makes it sooo easy right now").
>>
>>
>

Christian Szegedy wrote:
> View the leading dot as a compiler checked confirmation for the
> really thorough reader: "Yes, I really want an absolute path".
>
I think there's an echo in here.

http://www.nabble.com/Unfortunate-package-path-conflict-(security-vulnerability-)-tc19618678.html

-0xe1a

+1

On Thu, Apr 23, 2009 at 5:19 AM, Kris Nuttycombe
wrote:
> I've stated this before, but requiring a marker for a root-relative
> import is something that every Java developer who ever comes to the
> language will trip over. Probably multiple times. Relative imports are
> great, but it is *they* which should be prefixed.
>
> On Fri, Apr 17, 2009 at 1:47 AM, martin odersky wrote:
>> I think the tendency of the people here is indeed to pick leading `.' for root.
>>
>> Cheers
>>
>>  -- Martin
>>
>> On Fri, Apr 17, 2009 at 8:48 AM, Detering Dirk
>> wrote:
>>>> From: Ricky Clarkson [mailto:ricky.clarkson@gmail.com]
>>>> I taught Java to novices for a couple of years.  Most common
>>>> beginning mistake in writing Swing programs?
>>>>
>>>> import.javax.swing.JFrame;  In case it's not obvious, I'll
>>>> add some whitespace:
>>>> import .javax.swing.JFrame;
>>>>
>>>>
>>>> I wouldn't want to make this accident actually valid code. :)
>>>
>>> Why not?
>>> In Java it is a mistake simply because it's wrong.
>>>
>>> In any other language, if it is valid code, it is
>>> simply no mistake.
>>>
>>>> I should make a script that automatically replies to messages
>>>> mentioning Eclipse, Netbeans or IDEA, saying "does the
>>>> problem persist without the IDE?". :)
>>>
>>> :-)  +1 here
>>> (BTW: I wanted something like that in a Java list,
>>> where every improvent to Java The Language was answered
>>> with "Why? The IDE makes it sooo easy right now").
>>>
>>>
>>
>

On Thu, Apr 23, 2009 at 5:19 AM, Kris Nuttycombe
wrote:
> I've stated this before, but requiring a marker for a root-relative
> import is something that every Java developer who ever comes to the
> language will trip over. Probably multiple times. Relative imports are
> great, but it is *they* which should be prefixed.
>
One can argue back and forth about that. But the hard fact is,
changing that now is not an option because a lot of code would break.

Cheers