Wicked Cool Java

[Luke Webber]

E.g. that JavaComm was not maintained for years.

E.g. that JavaComm for Windows was recently withdrawn by Sun.

Ah, OK. I didn't realise that. I've used JavaComm in the past with some
success, but that was a few years back.

E.g. that JavaComm never made it into the standard edition.

E.g. that JavaComm is difficult to deploy via WebStart.

I should hope so. I wouldn't want you deploying DLLs to my system via
the web.

E.g. that JavaComm does not provide access to more esoteric tty options.

These problems.

Fair enough. That's why I asked. Thanks for enlightening me.

Luke

 

[Luke Webber]

I wrote serial drivers twelve to fourteen years ago, and today, I
honestly cannot tell you how I did it. I'm totally serious, I was doing
things in C and x86 asm that I literally do not know how to do today.
Some of the systems I worked on were pretty cool, but definitely not the
code or the process of writing it.

Heh. My first foray into serial comms was on a CP/M box in 1981,
starting in assembly language, but switching to Pascal. Then later in a
bizarre virtual assembler for the Pick O/S and some BASIC, then later in
x86 assembler for the IBM PC under DOS, then later in C for the same
platform, then later in C under Winblows and then later in Pascal under
Winblows. Too much to forget.

I don't remember anything about it,
and except when something reminds me (like this message thread), I
actually forget I had that job altogether.

I still recall fumbling around to learn just what the hell you could do
in a PC hardware interrupt routine. Not a lot, as it turned out. <g>

Luke

 

[Chris Uppal]

I should hope so. I wouldn't want you deploying DLLs to my system via
the web.

You can, though. I assume that it isn't possible to install /system/ DLLs via
WebStart, but you can certainly install application-local ones.

-- chris

 

[Oliver Wong]

Some of the reviews on the site are signed, it's up to the reviewer
whether they want to put their name to a review or not.

Out of curiosity, why would a reviewer NOT want to put their name on
their reviews?

- Oliver

 

[Dale King]

E.g. that JavaComm was not maintained for years.

E.g. that JavaComm for Windows was recently withdrawn by Sun.

E.g. that JavaComm never made it into the standard edition.

E.g. that JavaComm is difficult to deploy via WebStart.

E.g. that JavaComm does not provide access to more esoteric tty options.

These problems.

A couple more you forgot:

- Sun never made a JavaComm for Linux or Mac

- JavaComm was written to rely on a completely unnecessary properties
file that was found using stupid, undocumented, and obscure means
instead of using the more logical service provider abstraction
(<http://java.sun.com/j2se/1.3/docs/guide/jar/jar.html#Service Provider>).

I have told people for years to forget about JavaComm and use the open
source RXTX library which solves many of these problems
(http://www.rxtx.org).

 

[Dale King]

I should hope so. I wouldn't want you deploying DLLs to my system via
the web.

That's too narrow a perspective. In my case I had a monitoring device
for development that I created that talked serially to the PC. I wanted
to deploy the PC application for users of the device within the company
using WebStart and found it was quite a pain.

For more information about JavaComm and WebStart see:
- http://lopica.sourceforge.net/faq.html#comm

 

[Chris Uppal]

Out of curiosity, why would a reviewer NOT want to put their name on
their reviews?

To avoid reprisals ?

To preserve a "handle" by which s/he was already well-known in other contexts ?

To maintain a fixed identity, independent of [future] changes in his/her
professional or private life ?

Because "he" is actually a group of people ?

Because "they" are actually all the same person ?

And that's without getting into questionable or dishonest motivations...

-- chris

 

[Oliver Wong]

Out of curiosity, why would a reviewer NOT want to put their name on
their reviews?

To avoid reprisals ?

To preserve a "handle" by which s/he was already well-known in other
contexts ?

To maintain a fixed identity, independent of [future] changes in his/her
professional or private life ?

Because "he" is actually a group of people ?

Because "they" are actually all the same person ?

Right, sorry, I by "their name", I meant "a name". I'm familiar with the
practice of authors using pen-names which vary depending on the genre or
target audience they're writing for. You'd use one name for horror novels,
and another for romantic comedies, for example, so as not to have your
romantic-comedy fans stumble upon your horror novels and vice versa. I
suppose some sort of parallel could be drawn for reviews.

But my understanding is that some of the reviews on the site mentioned
upthreaded have no reviewer-name at all. The "avoid reprisals" seems to be
the only reason applicable for this context, but do reviewers really have to
fear reprisal? Unless you were giving a poor review for a product released
by the company you worked for, I really imagine getting into much trouble
unless the text you're submitting isn't so much a review, but more of a
whistleblowing for a dangerous, malfunctioning product, for example.

As Rhino said, the review is less convincing if the reviewer isn't
willing to stand by and defend his/her judgement.

- Oliver

 

[Chris Smith]

But my understanding is that some of the reviews on the site mentioned
upthreaded have no reviewer-name at all. The "avoid reprisals" seems to be
the only reason applicable for this context, but do reviewers really have to
fear reprisal?

The Mafia's gotta do something to survive! Who says they don't have
book publishers among their clientele?

(Seriously, I agree with you.)

--
www.designacourse.com
The Easiest Way To Train Anyone... Anywhere.

Chris Smith - Lead Software Developer/Technical Trainer
MindIQ Corporation

 

[steve]

I wrote serial drivers twelve to fourteen years ago, and today, I
honestly cannot tell you how I did it. I'm totally serious, I was doing
things in C and x86 asm that I literally do not know how to do today.
Some of the systems I worked on were pretty cool, but definitely not the
code or the process of writing it. I don't remember anything about it,
and except when something reminds me (like this message thread), I
actually forget I had that job altogether.

we ve all done this , long before C was ever invented. serial & parallel
printer drivers, computer to computer links before ethernet , more than 25
years ago.

Unfortunately i still have the code stuck in my head.

But my point remains that , "wicked cool java" is obviously a morons
Guide.
These kids , will never understand the true meaning of pain , or what is
actually required in real business environments.
(scanners/printers/cash draws /production line control/shipping/Quality
assurance and such like)

Hence RS232 & USB

Steve

 

[Stefan Ram]

we ve all done this , long before C was ever invented. serial
& parallel printer drivers, computer to computer links before
ethernet , more than 25 years ago.

Some time ago I meet with a friend. I had my Pet 2001 with me.
He had an Apple ][ clone. We connected them with a wire. Then,
we talked about a protocol. Within the next two hours or so we
both implemented it in 6502 machine language, he on his Apple
][ clone and I on my Pet 2001. Afterwards, we were able to
exchange data.

The Pet 2001 had an 8 bit userport. A loudspeaker could be
connected to a single bit (better with a small amplifier in
between). Someone figured out how to use this not only for
playing music, but also for generated speech output (via a
single bit!) using something that is called (IIRC) pulse-width
modulation.

Still, I can not follow the bashing on that book here. The
contents of the book is given in detail on its web site. If
one is interested in this contents, it should make sense to
get a copy of that book. The remarks about »RS232« and so are
not relevant in this regard. The book does not claim to
treat RS232, so it can not be blamed for not treating it.

 

[Thomas Weidenfeller]

A couple more you forgot:

- Sun never made a JavaComm for Linux or Mac

Sun now makes a JavaComm for Linux! Probably in exchange for dropping
the Windows version.

Of course, no one at Sun manages to update their 5+ years old, outdated,
artificial, toy JavaComm FAQ to mention this
http://java.sun.com/products/javacomm/reference/faqs/index.html That FAQ
still claims there is no Linux version.

- JavaComm was written to rely on a completely unnecessary properties
file

I thought I covered this with the remark about JavaComm being difficult
to deploy via WebStart It is that brain-dead file (not the native
libs, as others have suggested) which makes it a pain to deploy it via JWS.

I have told people for years to forget about JavaComm and use the open
source RXTX library which solves many of these problems
(http://www.rxtx.org).

And, which like everything else, creates new problems or has the same
problems. Last time I looked at it RXTX was also difficult to deploy it
via web-start. It e.g. required super-user rights to set the access
permissions on the serial devices on Linux.

/Thomas

 

[Chris Uppal]

But my understanding is that some of the reviews on the site mentioned
upthreaded have no reviewer-name at all. The "avoid reprisals" seems to be
the only reason applicable for this context, but do reviewers really have
to fear reprisal?

I admit that suggestion was a little frivolous. I imagine that if a publisher
were displeased with a review, then they might apply sanctions[*] to the
website, but hardly to the reviewer.

([*] No more free review books, no more advanced notices, no more invites to
publicity events. All of this on the (unlikely?) assumption that the
publisher is providing these goodies in the first place.)

Here's an imaginary scenario which I think could give rise to unidentified
(more than merely anonymous) reviews. I'm not suggesting that the techbooks
website's history resembles this in any way at all.

Some one (call him/her the editor) starts up a review website. At first all
the reviews are written by the same person, so there is no effective
distinction between the identity of the reviewer and the website itself (for
the purposes of "brand recognition"). I.e. all reviews are perfect reflections
of the editorial policy, and are merely written by an unidentified "staffer",
like many[**] newspaper articles. So there are no "names" on the reviews
themselves.

([**] But not enough. The cult of the reporter gets up my nose. Who /cares/
whether the reporter is John Middlethorpe or Amanda Dickinson. And it's even
worse on television news. Damned inane reporters gibbering platitudes at the
camera, and to cap it all insisting of telling us their names! Why do they
think we care ? Why are they wasting precious broadcasting time with their
personal problems ? Is it so that they may more easily be hunted down and
obliterated !? Anyway <cough/>, getting back to normal service...)

But it's easy to imagine that outsiders might supply reviews as the websites
reputation grows. Maybe even Big Names of the computer industry might do so (a
dream scenario for the editor ;-) If so, then they would probably want to
provide names, or at least identifiers. So, now we end up in a situation where
some, but by no means all, of the reviews are identified.

As Rhino said, the review is less convincing if the reviewer isn't
willing to stand by and defend his/her judgement.

BTW, my thanks to Rhino for the expression of confidence (some way up-thread).

-- chris

 

[Oliver Wong]

Here's an imaginary scenario which I think could give rise to unidentified
(more than merely anonymous) reviews. I'm not suggesting that the
techbooks
website's history resembles this in any way at all.

Some one (call him/her the editor) starts up a review website. At first
all
the reviews are written by the same person, so there is no effective
distinction between the identity of the reviewer and the website itself
(for
the purposes of "brand recognition"). I.e. all reviews are perfect
reflections
of the editorial policy, and are merely written by an unidentified
"staffer",
like many[**] newspaper articles. So there are no "names" on the reviews
themselves. [...]

But it's easy to imagine that outsiders might supply reviews as the
websites
reputation grows. Maybe even Big Names of the computer industry might do
so (a
dream scenario for the editor ;-) If so, then they would probably want
to
provide names, or at least identifiers. So, now we end up in a situation
where
some, but by no means all, of the reviews are identified.

Okay, makes sense. If I were the editor, when I'd modify the DB to
insert an "author" column, I'd have set the default value of the column to
myself, so that all existing articles are retroactively associated with me,
but maybe that's just me.

- Oliver

 

[Dale King]

Sun now makes a JavaComm for Linux! Probably in exchange for dropping
the Windows version.

Of course, no one at Sun manages to update their 5+ years old, outdated,
artificial, toy JavaComm FAQ to mention this
http://java.sun.com/products/javacomm/reference/faqs/index.html That FAQ
still claims there is no Linux version.

That's news to me.

I thought I covered this with the remark about JavaComm being difficult
to deploy via WebStart It is that brain-dead file (not the native
libs, as others have suggested) which makes it a pain to deploy it via JWS.

I agree that is why it is difficult to deploy via webstart, but the
properties file is also factor in using it outside of WebStart. It is an
extra file that they require copying.

I also wanted to highlight just how stupid their use of that proerty
file is. All that properties file has in it is a line like:

Driver=com.sun.comm.Win32Driver

All the API needs this for is to do a Class.forName of that class to
load the class and run initialize on it. You would think that they would
allow you to do that yourself (just like they do with JDBC drivers). But
for some unfathomable reason they were not satisfied with that. For some
reason, they wanted to make absolutely sure that the properties file
existed. Before most operations it makes a call to the security manager
asking if the file is deletable! How stupid is that?

And, which like everything else, creates new problems or has the same
problems. Last time I looked at it RXTX was also difficult to deploy it
via web-start. It e.g. required super-user rights to set the access
permissions on the serial devices on Linux.

That is more because of Linux and the default permissions in some
distributions than anything to do with RXTX or webstart. If the user
account does not have permissions to use the port there is nothing that
RXTX can do about that.

The thing that really bugs me is when JavaComm is held up as some form
of standard. Just because something comes from Sun does not make it a
standard, especially if they refuse to do any maintenance of it or
include it in the standard install. RXTX is really the standard and
there is no reason to install the JavaComm adapter on top of it.