Giving an application a window icon in a sensible way

[wesley.hall]

No misapprehension at all. Please consider reading the first paragraph

Please tell me why you think it should ever even occur to me to visit
this URL when not developing a weblication of any kind?

Perhaps you might have considered having a brief look before suggesting
that it was completely irrelevant to your project. You said this with
such conviction, yet you didn't even have a clue what it was. Again,
this reflects on your intellect.

What interview?

I couldn't have said it better myself!

 

[wesley.hall]

Most probably because Java WebStart has _absolutely_ *nothing* to do
with Web applications?

Yet again you judge the software features by its name (do you use Word
to write single-worded documents?)

....Or get upset because searching for 'word' in google links you to a
Microsoft word processor application and doesn't tell you anything
about the meaning of 'word'.

 

[Twisted]

E.G. <http://www.physci.org/pc/jtest.jnlp>

Sorry, I don't have any software on my system for interpreting .jnlp
files, whatever those are. (And I *do* have software for the common and
even many of the more obscure formats for images, archives, and the
like, just to put that into some sort of perspective...)

 

[Patricia Shanahan]

(e-mail address removed) wrote: ....

For a one-person project? You're joking. It would take me weeks to
learn to use such a complex tool, and then I have to operate some kind
of a server, then some kind of equally unfamiliar client ... that even
has *security* implications, since I have to make sure that the server
isn't visible to the outside world if I start running a server of some
sort.

....

My current project is a one-person project, part of my academic
research. I keep a subversion repository for it on a university server,
which provides both a revision history and an off-site backup. I use an
ssh tunnel so that the files in transfer to/from the server are encrypted.

If I didn't want an off-site backup, I would do revision control locally.

For one thing, it gives one great freedom to experiment. If I want to
try something out, I make sure I'm fully checked in, and do it. If it
doesn't work out, I just revert the previous version.

I realize that this may not be a good time to experiment with revision
control for you, but do not eliminate the idea just because you are
working on a one-person project. I don't think it would take you weeks
to learn.

Patricia

 

[wesley.hall]

Sorry, I don't have any software on my system for interpreting .jnlp
files

Ahhh, I think I am beginning to see the problem...

 

[Thomas Kellerer]

> For one thing, it gives one great freedom to experiment. If I want to
try something out, I make sure I'm fully checked in, and do it. If it
doesn't work out, I just revert the previous version.

I realize that this may not be a good time to experiment with revision
control for you, but do not eliminate the idea just because you are
working on a one-person project. I don't think it would take you weeks
to learn.

I couldn't agree more!

It took me about 10 minutes to figure out how to use a local CVS
repository...

Thomas

 

[Twisted]

Daniel Dyer wrote:
[snip some opinions on what "should have been obvious" and the like]

This should do the job. Save the following as build.xml in the root of
your project. Edit the properties at the top to match the paths in your
project, and then run "ant" or "ant all" from the command line (from the
root directory of your project). I don't remember how you set Eclipse up
to use this, but it's trivial in IDEA and NetBeans, so I can't imagine
that it's very difficult. If you need to extend this script, for example
to add a manifest, I'm sure you can work out how from the examples in the
manual (http://ant.apache.org/manual/index.html).

Given that Eclipse already rebuilds things automatically when they have
changed, what advantage does this provide? (That's an honest question,
not a rejection, disparagement, or anything else you may be tempted to
misinterpret it as, by the way.)

 

[Twisted]

Running the server portion of a version control system locally just
means you have a bunch of files that maintain all changes to your
software. It is no less secure than your current approach, you just
have a few more files in a specific format.

You seem to be forgetting the cardinal rule of system administration:
because way, way too many dolts ship server software installation
packages whose default install configuration is "wide open to the
Internet, with a widely known default password" it is important to take
extra precautions when installing any kind of server software to
configure it correctly and block anything via your firewall that you
aren't sure about. (If you find you need it later you unblock it then.
And change that default password!)

Even if the "default password" hole doesn't get you, an exploit of some
kind may well do so (recall the Slammer worm -- everyone with an
unfirewalled database got hit, and I do mean "everyone", within ten
minutes of the thing being released!) so firewalling away any server
functionality that doesn't need wide-area access is a crucial (and
oft-neglected) security precaution.

You on the other hand seem to be treating the installation of a version
control server as no more an issue in regard to security than
installing a new word processor or spreadsheet.

This is exactly what I mean when I say, "Stop assuming you are the
smartest person around". Why assume that everyone who runs their
software this way is running insecurely and hasn't even realized.

I'm not. You're assuming I'm assuming that. When all I actually said
was that installing a server introduces new security considerations
that are absent when installing other software. If you dispute THAT
statement, I pray that you aren't a system administrator anywhere that
keeps anything working that I happen to use or otherwise have reason to
care about.

And to pre-empt your predictable response to what I said above -- I at
no time claimed that any particular version control product ships "wide
open with a default password", only that there is an awful lot of
server software (and hardware!) that does, and consequently a savvy
sysadmin has to treat *all* servers as if they *might* arrive in such a
state and make sure to immediately configure them (and their firewall)
properly. It adds to the overhead -- especially if you get it wrong and
get exploited.

The fact that you 'tell' people about the insecurity rather than post,
"Ok, that is interesting, but wouldn't running a server leave me
susceptable to remote attacks?", is why people are taking you for and
treating you like a moron.

I *did* say the equivalent of "Ok, that is interesting, but wouldn't
running a server leave me susceptible to remote attacks?". You are
putting words in my mouth when you suggest anything different. The
exact thing I said was closer to "...and installing server software
also adds *security* considerations..." -- not anything about it
invariably reducing your security.

If you can't even attack the things I genuinely said and feel the need
to make it up as you go along and attack the resulting straw men
instead, then we're done here and you can shut up now.

Lack of industry experience is not the reason you are very unlikely to
make it as a professional developer.

No, the glut is, combined with the likelihood that the would-be bosses
are nearly all assholes like you. It's been my observation that shit
floats to the top, and that in actual real-world management situations
it's usually some know-it-all prick with an attitude problem like you
who ends up telling the actual technical staff how to do their jobs,
and then sending heads rolling when they do exactly as instructed
against their own best judgment and then everything crashes. Currently,
this seems to have happened at infoworld.com, one of whose hosted blogs
has been on the blink for over 12 hours straight now, no doubt due to a
stroke of massive incompetence. I've never seen incompetence of that
magnitude that wasn't wearing a suit and a tie and drawing an
upper-five-figure salary, ever in my (considerable) life. No doubt a
manager there did something dumb, like tell the techs how to do their
jobs until the system was riddled with bugs and required 24/7
nursemaiding to keep running, and then fire all but one of them to save
money and let the remaining one go on vacation or call in sick with
no-one to cover for him...exit one formerly-functioning server and
Christ alone knows how many prospective customers. The odd thing is
that these guys have doddered along for decades without all wiping
themselves out in a giant stock market crash. I mean, there should have
been at least ONE more Great Depression with bozos like those running
the show, you'd think. Maybe enough technicians stick to their guns and
actually do their jobs properly despite their management that
everything just keeps on muddling along. Still, every guy like you to
come along with an attitude like yours only matched by a comparably
poor grasp of the basic principles of logic makes the likelihood of a
civilization-destroying disaster of comparable magnitude happening
tomorrow notch up just a tad. And there are an awful lot of you, and it
all adds up...

 

[Daniel Dyer]

Given that Eclipse already rebuilds things automatically when they have
changed, what advantage does this provide? (That's an honest question,
not a rejection, disparagement, or anything else you may be tempted to
misinterpret it as, by the way.)

Well, I was under the impression that you hadn't been able to get Eclipse
to build the JAR file for you as part of an atomic, automated build. If
you can get Eclipse to do this for you, then maybe you don't need Ant at
this stage. The suggestion was also based on the assumption that building
a JAR file was the "right" solution for your problem. This is not
necessarily the case, but it is the solution I would prefer in your
position.

The main reason that I like to use Ant for the build rather than the IDE's
built-in mechanism is because it stops you from being tied to the IDE.
This has two advantages, which like many of the suggestions you have
received on this thread may not be so important for a one-person team.
The first advantage is that different people working on your project can
use different IDEs or editors but still have a single "official" way of
building the code (if you try to configure multiple tools to do the build,
you may introduce subtle differences).

The second advantage is that you can build from the command line without
requiring any graphical tools. This is advantageous for integrating with
other processes, such as continuous integration systems. It is also
better if you are planning to distribute your application in source form,
since it will be easier for other people to build the software. They
won't have to install Eclipse or work out all the dependencies for
themselves. Of course, you don't have to use Ant for this, you could just
write a shell script or batch file. The advantage of Ant is that it will
run on any system that has a JVM and that it has built-in support for the
most common things you would want the build process to do.

Dan.

 

[Twisted]

Yes, the picture I get: you have no idea how to seupt a software
development environment.
The picture I get is: you have never heard of batch files, not to speak
of build tools and the concept of an IDE does not seem to be understood
by you either (sorry if that sounds insulting, but this *is* the picture
that you communicate)

Actually, the picture that *you* communicate is one of someone who,
despite not having read more than a fraction of the specific post
you're following up to nor a single word of the rest of the thread,
nonetheless has a strong opinion and feels the need to publicize it. Of
course, it turns out to be one of those opinions that more properly
belongs deposited in a certain ceramic vessel and then flushed, with
the smell to prove it...

I simply press F6 to run my application and *everything* is
automatically taken care of.

With my *current* setup, it's quite similar. Until I go doing something
that means having to copy something manually regularly as well, even if
I automate that down to one additional keypress or mouse click.

I don't know Eclipse, but as others have pointed out repeatedly, Eclipse
can be setup in just the same way.

And five minutes ago you claimed that I was "obviously" not using an
IDE, or even aware of what one was. Curious.

But please stop asking questions around here if you don't like the answers.

I haven't passed much judgment on the answers, unless you consider
asking for additional information to constitute disparagement all on
its own.

It's certain peoples' tone I don't like.

 

[Twisted]

Most probably because Java WebStart has _absolutely_ *nothing* to do
with Web applications?

My point apparently sailed right over your head and into the next
county. Because you sort-of made it for me just now, yet you seem to
think you just scored a point *against* me.

If what you say is true, but a particular person *does not know that*,
and they are looking for particular *functionality* (not net related)
without knowing the *name* you mention, then it is unlikely that they
will ever look to that URL for their answers. They'll think "Hrm, looks
like networking stuff" and move on to the next link. Yet you assume
that they'll look there and find out whatever else it does *despite*
its looking irrelevant.

You have a chicken and egg problem. It looks irrelevant until you've
been there, but you won't go there until it looks relevant ...

Perhaps now you understand? And perhaps you understand why I don't care
much for the assumptions people are making about how I should already
know this or already have seen that. Maybe in an ideal world I should
have, but it's unreasonable to expect me to have actually done so if
they are, for whatever reason, in effect cleverly disguised as
something irrelevant to my goals.

But then, I don't think people like you are fazed much by the notion
that some of your expectations might be unreasonable; you just merrily
hold them anyway, and then fault people for not satisfying your
expectations, even when those people have been perfectly honest and
acted in good faith.

Yet again you judge the software features by its name (do you use Word
to write single-worded documents?)

It at least has something to do with words. If something with "web" in
its name has nothing to do with webs (world wide or otherwise) then
there is a problem.

 

[Twisted]

Perhaps you might have considered having a brief look before suggesting
that it was completely irrelevant to your project.

I never suggested that it was; merely that it appeared to be. If it
isn't but appears to be, then that in itself may constitute a problem
in its own right, but my not "considering having a brief look" is not,
any more than my not looking at the front cover of today's New York
Times for the answer is any sort of failure of due diligence on my
part.

You said this with
such conviction, yet you didn't even have a clue what it was. Again,
this reflects on your intellect.

Here's exactly what I said:

You interpret this as "Please tell me why you think it should ever even
occur to me to visit this page, with these contents, when not
developing a weblication of any kind?" as if I know (or should know)
something about the contents already, and was referring to those. I
actually meant, and this seems to be the more logical interpretation of
what I actually wrote by the way:

"Please tell me why you think it should ever even occur to me to visit
a URL with this name when not developing a weblication of any kind?"

Note the difference; and that the only reference is to the URL's name.
If something relevant is there, but nothing about *the URL itself*
suggests that that is even fairly likely, then why, indeed, would
*anyone* follow it (given those particular goals, with regard to which
it doesn't look all that relevant)?

Until you can answer that question to my satisfaction, we're done here.

I couldn't have said it better myself!

Hello? Yes. Sell everything else and put it all into gold and canned
foodstuffs. Yes, you heard me. Yes, I did mean "everything" when I said
"everything"...

 

[wesley.hall]

You seem to be forgetting the cardinal rule of system administration:
because way, way too many dolts ship server software installation
packages whose default install configuration is "wide open to the
Internet, with a widely known default password"

Please read these words and read them as carefully as you can, trying
very very hard to understand....

Installing a local version control system does not run any kind of
server process, of any kind, anywhere. There are no default passwords,
no back doors, no additional security risk. It is just a bunch of
files, on your disk, that keep track of your incremental software
changes. I simply cannot make it any clearer than that, if you still do
not understand then there is no hope.

If you eventually decide to use a remote server setup, then subversion
is accessible via an extension to the HTTP protocol called DAV. Your
subversion server is generally an apache server, which is widely used
and well maintained. In this configuration, you would have security
considerations, but nowhere near the magnitute you are suggesting. I
run a linux server with my subversion server, it is not difficult to
install, configure or secure, all processes are well documented. There
is a theorectical risk of compromise with any remotely accessible
server, but an apache compromise would be huge huge news and a patch
would be available in minutes (it would, after all, affect the majority
of all webservers). Not running these standard tools due to attack
paranoia is akin to not leaving your house for fear of being hit by a
bus.

Of course, you didn't know that subversion generally uses apache as
it's server, but you probably didn't even bother to try to find out for
yourself either.

The rest of your post is pure gold. I am an asshole, a
counter-productive manager, and the potential destroyer of civilization
as we know it. At least I have some references for my resume now.

 

[Twisted]

My current project is a one-person project, part of my academic
research. I keep a subversion repository for it on a university server,
which provides both a revision history and an off-site backup. I use an
ssh tunnel so that the files in transfer to/from the server are encrypted.

And you probably wouldn't if the following weren't true:
a) You were already familiar with this sort of thing, as a result of
larger projects;
b) All of this stuff, including the hosting, you were not paying extra
for;
c) You're familiar enough with setting up things like ssh tunnels and
security precautions to trust yourself to do this stuff without
screwing it up and letting every Tom, Dick, and Harry with a rootkit
and subscription to 2600 get in by mistake.


I can't actually state anything about what you would or wouldn't do
with certainty, though; it's entirely possible that you happen to be
atypical in this area anyway.

I realize that this may not be a good time to experiment with revision
control for you, but do not eliminate the idea just because you are
working on a one-person project. I don't think it would take you weeks
to learn.

I never said I was "eliminating the idea", only questioning its being
anything but overkill and added work for little gain in the context of
this specific project. But then, that seems to be a recurring
misunderstanding around here, where my not jumping up all enthused and
immediately going and doing something is interpreted as a permanent
rejection of that thing for some reason, or at least my asking a
question is interpreted as a permanent rejection. (I would expect it
should be more indicative of some sort of (possibly guarded) interest;
outright rejection would probably not involve questions at all, other
than the rhetorical variety, and most likely would involve either
blanket unconditional flat refusals or just silence.)

 

[nebulous99]

I couldn't agree more!

It took me about 10 minutes to figure out how to use a local CVS
repository...

I find that difficult to believe, unless the loophole is the obvious
and you already knew a great deal about how to use a *remote* CVS
repository, which is probably more complex.

10 minutes is the time to figure out how to get an unfamiliar media
player to import your playlist, shuffle, and loop, or to go from IE to
Firefox or something.
1 hour is the time to figure out something more complex, such as doing
some basic stuff with a spreadsheet with no or only sporadic prior
spreadsheet experience and no familiarity with the specific software
used, or get a "hello world" working in a new programming language
(actually that can be under 10 minutes if it's fairly simple to use --
BASIC, Smalltalk, some Lisps; generally longer for anything that needs
various tools installed and configured correctly, e.g. most C or C++
environments, with no prior experience with the tools or language --
most of the time spent getting the build to work right).

Then there's stuff like advanced spreadsheet or CAS functionality, or
3D modeling (hours to get started, and as much as one to learn new
software with proficiency with the general category). And then there's:
* Configuring and installing a server yourself (even intended for a
single user).
* Configuring and installing the client.
* Reading their manuals, at least for the most basic and common how-to
operations plus the installation and configuration related stuff.
* Hammering out the kinks, which complex client/server architecture
software usually has more than zero of before it gets going properly on
a particular equipment configuration.
* Actually setting up a specific project on the system, once the thing
is running.
* Hammering out any problems that didn't manifest until you tried to
actually use the system for anything, rather than just turn it on.
* Learning, and getting used to, the new procedure for getting your
data out and putting it back in once you've made your changes.

One thing people already proficient with a tool often forget when
estimating the difficulty for new users is the amount of time and
mental effort involved in that last bit -- *getting used to* a changed
workflow.

I flatly disbelieve your ten-minute figure, unless you found a magical
CVS growing on a tree somewhere that you just unzip in a clearly
explained place on your hard drive and then some IDE you already knew
how to use (say, Eclipse) starts transparently and automatically using
it as a backend without further ado.

The odd thing is, I'm not entirely convinced that such a magical CVS is
impossible. It just sounds improbable, given the low expectations one
has regarding the ease of use of unfamiliar or substantially changed
software these days.

P.S. it's me, but replying with a different address. It seems that one
of you did something to try to block me from being able to respond to
the crap some of you are writing and Google is now claiming I've
violated some kind of limit. Of course they helpfully provide me a link
labeled "if you feel this message is in error ..." which leads to a
bare-bones 404 page.

Obviously, it didn't work, since I am replying now to let whoever you
are know that a) you failed to shut me up, b) you succeeded in pissing
me off, and c) you are now being called on it, as you should really
have been expecting, using such underhanded tactics.

I do not find this amusing, and I certainly do not consider such
tactics to try to win an argument to be anything other than cheating in
the first degree. If I identify the specific person responsible, there
will be consequences. I can only assume they either a) made a spurious
complaint of some nonexistent misconduct on my part or b) actually
acted more directly and hacked Google or otherwise messed with my
ability to reply. The latter actually is fairly plausible because it's
happened before -- some guy put some weird shit in the headers of his
posts so that every attempt to follow up to the drivel he spouted in
comp.os.windows using Google Groups choked with a "no such group" error
despite the fact that all his hapless victim had done was read his
drivel, immediately think of 15,000 objections, click "reply", type in
about the top 15 of the objections, and hit "post", which obviously
should Just Work(tm), but for this particular bonehead's posts
invariably failed. Manually editing some of the headers of the reply
(not easy on google's interface) made it work. I've also encountered
posts the replies to which would fail silently, failing to appear even
after several whole hours without actually producing any error
messages, either. I don't doubt there's some stupid script-kiddie trick
to make a post that will cause a GG user to be locked out of his
account if he tries to reply. It looks like the other tricks are based
on making GG redirect your posting away from the group you were in, in
the one case to a bogus group and in the other to an actually-existing
one you probably don't read. It follows that such a trick can also be
used to redirect the posting to where it will be interpreted as spam
and generate complaints, or perhaps even to make it generate a large
number of copies that makes GG's computers think you're a spammer or
that rapidly exhaust some kind of limit meant to block bot postings
that a human shouldn't actually be able to reach in normal usage.

It doesn't actually matter *what* was done, only that if anyone ever
does anything like that to me again, then it will be the last time they
ever do anything to me, period. I *will* find out who the guilty party
is, especially if there is *ever* a repeat of this attack, and I *will*
find out other things about them once I find out their name. Trust me
on this.

Capiche?

Now, I expect that in the future, hitting "reply", typing stuff, and
hitting "post message" will never again behave in any manner that is
surprising, and most especially will never again reject a posting of
mine. My postings do not ever deserve any kind of automated rejection
and I will take each and every occurrence of such as a direct and
mortal insult -- it's tantamount to accusing me of being a spammer, and
moreover, in the most indirect and cowardly of manners rather than just
coming right out and saying it to my face. Whosoever does such a thing
to me is only one rung above an *actual* spammer on the evolutionary
ladder and I will make certain to remind them of this fact on every
occasion that seems warranted, particularly when there happens to be an
audience!

And of course it's worth noting that no message to an unmoderated
usenet group should be blocked from being posted anyway. EVER.
After-the-fact cancellation (by the author, or if it's outright spam)
is ALL that is permissible here.

So don't let me EVER catch ANY of you trying to forcibly muzzle rather
than debate your opponent EVER AGAIN. (That includes any cases I become
aware of in which your opponent is somebody other than me, by the way;
fair's fair.)

Short version: Don't you EVER do that again! NAUGHTY, naughty boy! You
know who you are!

 

[nebulous99]

Well, I was under the impression that you hadn't been able to get Eclipse
to build the JAR file for you as part of an atomic, automated build.

Technically correct, but for lack of trying -- packaging for
distribution is not really even on the horizon here yet.

P.S. it's me. Some jerkwad, probably in this group, tried to forcibly
muzzle me when they ran out of (il)logical arguments to debate me with.
They succeeded only in muzzling "twisted" and earning a stern warning
that if they do the same thing to "nebulous" they will have a) still
not succeeded in shutting me up and b) made themselves strong
contenders for a near-future Darwin Award.

The main reason that I like to use Ant for the build rather than the IDE's
built-in mechanism is because it stops you from being tied to the IDE.
This has two advantages, which like many of the suggestions you have
received on this thread may not be so important for a one-person team.
The first advantage is that different people working on your project can
use different IDEs or editors but still have a single "official" way of
building the code (if you try to configure multiple tools to do the build,
you may introduce subtle differences).

Clearly that makes it important for multiperson projects, or as a
bridge during an IDE change for a single user, at minimum.

The second advantage is that you can build from the command line without
requiring any graphical tools.

Usually I only would need to build after changing something, which
would happen in the IDE.

It is also better if you are planning to distribute your application in source form

Ant also emerges as distinctly useful when distribution time comes and
the project is open source, then...

 

[Daniel Dyer]

P.S. it's me, but replying with a different address. It seems that one
of you did something to try to block me from being able to respond to
the crap some of you are writing and Google is now claiming I've
violated some kind of limit. Of course they helpfully provide me a link
labeled "if you feel this message is in error ..." which leads to a
bare-bones 404 page.

http://groups.google.com/support/bin/answer.py?answer=15331&topic=250

Dan.

 

[wesley.hall]

Note the difference; and that the only reference is to the URL's name.
If something relevant is there, but nothing about *the URL itself*
suggests that that is even fairly likely, then why, indeed, would
*anyone* follow it (given those particular goals, with regard to which
it doesn't look all that relevant)?

Ok, lets recap....

* You asked about including an image with an application (remember
that?).

* You received the suggestion that you should put the image in a jar
file and use getResource

* You stated that your way was better because building a jar file was a
problem/hard/time consuming

* I suggested that you might want to use ant as this would make the
process very easy

* You asked why would I want to use ant when I can use eclipse

* I gave you a list of many things you could do with ant that were
difficult or impossible with eclipse, some where relevant to your
current project, some weren't but the point was, lots of extra things
were possible

* You picked one item on the list that didn't seem (to you) to be
relevant to your project and attempted to use it to claim that I didn't
understand the problem.

* I pointed out that it was infact relevant, you had just
misinterpreted the name. I did this very politely dispite being a
little annoyed that you would rather just blindly claim I didn't
comprehend your problem rather than do a very small amount of legwork
to be sure I didn't.

* You started moaning about the name of the project and "why would you
think to go there etc"

So... to answer your question...

No reason why you should go there unless your saw the mention of it in
my original post and had a sudden urge to look it up (and any decent
developer I have ever met WOULD, just because of curiosity), but there
is also no reason for you to start claiming that it was not relevant to
your project just based on the name alone. Infact, I am certain the
only reason that you are trying to blame the project name is because
you were caught out trying to be clever in telling me web start wasn't
relevant and are just nursing your injured pride.

 

[Bent C Dalager]

Sorry, I don't have any software on my system for interpreting .jnlp
files, whatever those are. (And I *do* have software for the common and
even many of the more obscure formats for images, archives, and the
like, just to put that into some sort of perspective...)

You actually don't have a web browser?

(I am assuming you have a JRE since you're apparantly doing Java
development.)

Cheers
Bent D

 

[Ian Wilson]

That's it?

That is what I said.

And it will be found by getResource during testing *and*
after deployment?

That is what I said.

(I assume using a relative URI like
getResource("resources/foo.gif")?

No, I didn't say *that*, *two* *days* *ago* I posted this:

> I use Eclipse, I have an image for my app in a subdirectory of my
> project. I have this code (and this code only) to load the image:
>
> setIconImage(new ImageIcon(MainForm.class
> .getResource("/resources/logo32.png")).getImage());

Note the leading "/".

Ring any bells?