[ANN] Atox 0.2 released

Richard James rmb25612 at yahoo.com
Sun Feb 15 12:06:58 CET 2004


Istvan Albert <ZZZ-ialbert-ZZZ at mailblocks-ZZZ.com-ZZZ> wrote in message news:<yfidnROF5MbLJLPdRVn-ug at giganews.com>...
> Richard James wrote:
> 
> > But, I wasted a lot of valuable time and credibility trying to sell
> > NOGUI uh... ANYGUI for in house rapid MS Windows gui development to my
> > boss and co-workers.
> > 
> > I don't know how many times I said: I'm sure the next release is just
> > around the corner. And it will run on ANY gui that our customers have
> > installed!!!!!
> 
> The question that pops into one's mind here is why would you promise
> things that are completely out of your control? I understand your
> frustration with unfinished or dead projects but let's remember that
> for each successful project there will be hundreds that die off and there
> is nothing wrong with that. Survival of the fittest.
> 
> i.

Well it was part of my strategy of getting Python accepted into the
work place as a serious work tool. The early versions of ANYGUI were
very promising in the simplification of GUI interfaces. The command
line interface may be second nature in the Linux hacker world. But in
the "real" MS Windows business world it is not a "production level"
interface.

The true power of Python is in being able to do tasks, like lists and
dictionaries, without thinking about it very much, and without
fighting wild C pointers, or engaging in "in house" C++ class warfare
over what type of list classes and or structures we use this week.

Python is productive, but Python guis are a weak point, as is
compilation.

(Now really! How different are interpreters and linkers? Other than
everybody loves to design and write interpreters and few have ever
written a linker. My "FREE radical" idea is to define and generate
bytecodes that can be both interpreted and linked. Another one is to
dump all the C++ gui paradigms and get the assembly language game
programmers involved in creating a FAST gui programming language you
can use without spending 4 years reading the unclear out of date
documentation. Does QT really need 5000 plus widgets? Or would a
general purpose widget design language be better? I know gui
programming keeps many gui gurus employed, but doesn't "survival of
the fittest" mean constant challenge of the old guard? Flame away at
me. But the Emperors in these two programming areas have no clothes.
You can only wrap so many abstraction layers on top of interface
layers before you strangle yourself. And don't get me started on the
bizzare flavor of the week internet browser html flash asp js java
notjava C# mess!)

Programming productivity matters, the faster you can produce a
finished Windows program the happier the customer is. And you get all
those widgets you buy delivered to the store on time. Instead of half
built widgets laying on a stalled production line, while the packaging
machine GUI is locked up over night.

The academic side of the Python world may not appreciate the "time is
money" concerns. But if time didn't matter, I would write Windows apps
in X86 Assembly and get 100X perfomance increases over the slow
wxPython dll.

There still are a lot of older slower X86 machines running 24/7 on the
manufacturing floor. You don't need a 4GHZ machine to stamp date
codes. Waiting 2 minutes for wxPython to load while the boss stares at
the back of your neck makes Python look very bad. Gui4Cli pops right
up and can talk back and forth to other G4C guis and even other native
Windows apps that are running in the background.

And adding a custom gui for a monitoring program, to alert production
operators to catch a flakey machine in the act, is now childs play
instead of a 40 hour C++ gui programming nightmare weekend.

So I gambled on ANYGUI and lost, up until that point, the Python
community had not let me down. ANYGUI had a snazy graphic on it's
webpage and an active user and developers list going. And many people
seemed to be excitedly involved at the time.

More than anything, the book notice posted on the ANYGUI site in
August 2002 sealed the deal that ANYGUI was a "REAL" product.

I showed the "Practical Python" book to my boss. He liked the look of
it. Can I get a refund? At least rename it "Mythical Python". And
there has been NO notice on the website that the ANYGUI project has
been abandoned, just a "stay tuned" notice. The developers list seems
to have been abandoned by it's developers in August 2003.

Doesn't this "current" 2004 advert look like the "fittest" having
survived???
----------
>From Barnes and Noble 
----------
Practical Python is an indispensable guide for developers intent on
learning what they need to know to master the Python language.

- Author is well known for his online Python tutorials, Instant Python
and Instant Hacking, and is the founder and lead developer of Anygui,
a respected Python GUI unification project

- Ruthlessly focused on providing practical Python instruction,
foregoing those topics not of interest to mainstream developers
-----------

"a respected Python GUI unification project"???

Was all this just smoke and mirrors to sell a $50 book?

Is XML just the next hot book topic?

As a contrasting example, the pycurl webpage looked like a twenty
minute hand coded html job, yet pycurl has proven to be a reliable and
very useful aspect of the Python programming experience.

You can't judge a book by it's cover or a project by it's webpage
design or even a project by it being in an authoritative looking book.
So that is why I suggested some type of offical ranking page for
Python projects. Maybe others can avoid wasting time and effort in the
future.

The only consequence was that we were stuck with the SOW (Same Old
Way) method for a few more months. Until we found the FREE and well
documented Gui4Cli. We can now make simple gui changes in the Gui4Cli
script language in under half an hour. Most GUIs are also under 20K
bytes in size and using pycurl you can update the production floor
machine in seconds. For low to medium gui complexity Python with
Gui4Cli gets the job done. And it helps in testing out ideas that can
be later turned into full power C++ guis to handle higher data
collection and control rates.

So it turned out ANYGUI's failure to thrive, was as Martha Stewart
says: "a very good thing".

And now we almost don't know how we ever did things before we had
Python and Gui4Cli!

-- R.J.



More information about the Python-list mailing list