Python 'Lets Me See The Forest'

[Kamilche]

Man, I've been banging my head against a C program for a while now.
I'm embarrassed to admit how long. I really want to use Python, as I
mentioned in a prior post, but the speed hit is such that I'll
probably use it only for prototyping.

But boy, development sure is fast in Python! Today, while playing with
it, I thought of a better way to arrange my data, that makes the
program more flexible. It's a good enough idea to warrant redoing my
code (even my C code) to take advantage of it. When I went to write up
the pseudocode to process the new method, I found myself naturally
using Python.

C is just so... detail oriented. By the time I set up the data
structures, do the string manipulation and memory management, I can
hardly remember what it was I was initially trying to accomplish, much
less think of a better way to do it! Maybe I'm just getting old... but
Python just fits my idea of 'mental pseudocode' so well, it's hard to
resist using it. Dictionaries ARE an underutilized concept in DP, and
I was using them up the yin yang, even in my C program. Python meshes
well with my programming style.

--Kamilche

 

[Larry Bates]

I have a theory that the more experienced a programmer
is (e.g. has used a number of "hard" languages for
many years) the more they will like and be able to
use the power of Python. I've been programming for
over 30 years and I LOVE Python. I committed to doing
all my coding in Python about 2 years ago and have
never regretted it. I find that I'm still turning
up "jewels" of programming methodology every day.

The power of any good programming language is the
ability to "hide" complexity without limiting
flexibility. Python does an excellent job of this.
I wouldn't have any problem starting beginning
programmers with Python (it's much better than VB
for the pure beginner). The beauty is that unlike
many "beginner" languages it never seems to run
out of power. You can write incredibly complex
programs in Python that run well and are a joy to
maintain. Python is pseudo-code that actually
runs without being translated into some other
language!

The other thing I like is that it runs everywhere.
I've written Windows NT services, Windows NT COM+
objects, regular text mode programs, Windows GUI
programs (with wxWindows), Linux programs, dynamic
webpage programs, etc. Without Python I would have
to switch between 3-4 languages and would never get
really proficient with any of them.

The reusability of modules means that I can use
modules that I've written for one project and easily
use them for other projects. I start out with
thousands of lines of mature, debugged code in these
modules that gives me a real running start at the
project at hand. Now I know that this ability is
not unique to Python, it just seems that it gets
used more than in most other languages.

Glad you hare having fun.
Larry Bates
Syscon, Inc.

 

[beliavsky]

Man, I've been banging my head against a C program for a while now.
I'm embarrassed to admit how long. I really want to use Python, as I
mentioned in a prior post, but the speed hit is such that I'll
probably use it only for prototyping.

But boy, development sure is fast in Python! Today, while playing with
it, I thought of a better way to arrange my data, that makes the
program more flexible. It's a good enough idea to warrant redoing my
code (even my C code) to take advantage of it. When I went to write up
the pseudocode to process the new method, I found myself naturally
using Python.

C is just so... detail oriented. By the time I set up the data
structures, do the string manipulation and memory management, I can
hardly remember what it was I was initially trying to accomplish, much
less think of a better way to do it! Maybe I'm just getting old... but
Python just fits my idea of 'mental pseudocode' so well, it's hard to
resist using it. Dictionaries ARE an underutilized concept in DP, and
I was using them up the yin yang, even in my C program. Python meshes
well with my programming style.

--Kamilche

I am no C++ expert, to put it mildly, but couldn't some of your problems
have been solved by using C++ instead of C? C++ has data structures like
vectors, lists, and maps, with many algorithms in the STL to work with them.
Using references to pass arguments, one can avoid low-level pointers in many
cases. C++ strings are easier to use the C char's. There are templated classes
to replicate some of the functionality of Python's Numeric/Numarray or Fortran
90/95 arrays. C is more of a low-level systems language than an application
programming language -- comparing it to Python seems unfair to me, when the
choice of C++ exists.

For strictly numerical tasks Fortran 95 is in my experience both more readable
than Python (because of variable declarations and the ability to define constants)
AND much faster. You do not have to suffer Python's performance hit to program
in a clean, high-level language.

(I like Python, but there is so much pro-Python propaganda in this newsgroup
that some anti-Python messages may be a healthy corrective.)

 

[Kamilche]

I am no C++ expert, to put it mildly, but couldn't some of your problems
have been solved by using C++ instead of C?

Uh. Yeah. I know C++ sort of, I redid all my base code in preparation
for using it... and it decidedly DIDN'T mesh with my programming
style! :-O I gave it up after finding one too many complexities and
oddities in the language.

I've used quite a few languages... C, C++, Fortran, Pascal, Assembler,
COBOL, Basic, Visual Basic (6), Python, Javascript, and Hypercard. The
ones I used most heavily were COBOL, HyperCard, Visual Basic, and C.
The top 4 faves were VB, Python, HyperCard, and C, in that order.
HyperCard is dead now, but it had its day. Of the remaining 3, I'm
mainly using Python and C, since MS killed VB6. They all had their
glories. Note the marked lack of C++ in the top 4. :-D You have to be
superhuman, or maybe just supertwisted, to be productive in that
language. :-O

--Kamilche

 

[Kamilche]

... I've been programming for
over 30 years and I LOVE Python. I committed to doing
all my coding in Python about 2 years ago and have
never regretted it. I find that I'm still turning
up "jewels" of programming methodology every day.

Wow, you must be quite skilled in the intricacies of Python, then. Do
you have a web page where you share pearls of Python wisdom? From my
Google searches, it appears pages like that are still very necessary
in this language.

I wish I had a better book on it, or the online reference was more
comprehensive.

--Kamilche

 

[David Bolen]

I am no C++ expert, to put it mildly, but couldn't some of your
problems have been solved by using C++ instead of C? C++ has data
structures like vectors, lists, and maps, with many algorithms in
the STL to work with them. Using references to pass arguments, one
can avoid low-level pointers in many cases. C++ strings are easier
to use the C char's. There are templated classes to replicate some
of the functionality of Python's Numeric/Numarray or Fortran 90/95
arrays. C is more of a low-level systems language than an
application programming language -- comparing it to Python seems
unfair to me, when the choice of C++ exists.

I don't know if it's just me, but a funny thing I've found is that
after using Python almost extensively for the past 4-5 years, I find
it much easier to grasp and appreciate the C++ STL than I did prior to
Python. Of course, to be fair back before Python when I was using C++
more significantly, I hadn't really been out of the C world for too
long, and I recall compiler template support being more hit and miss
so trusting the STL (if it was even available) was tricky, but I think
Python helped get my mindset more acclimated to the higher level data
constructs, making it more natural to use their C++ counterparts when
working nowadays with C++, things which I tended not to utilize as
effectively in the past.

With that said, doing similar work in C++ versus Python even today
feels like I'm working in a tar pit while working on the C++ side.
Oh, I like the performance and part of me still can't help but enjoy
the feeling of working closer to the metal, but egads, the extra
typing (particularly with the STL), the fighting with the compiler's
type checking, those horrible template error messages. Heck, basic
iteration, or even the hoops to jump through to bind instance members
for callbacks. The STL is a valuable set of generic containers and
algorithms that I wouldn't want to do without in C++, but similiar to
the original poster's point, I still feel burdened down with details
when working with them compared to equivalent Python code, even when
the code is algorithmically and data structurally similar.

Of course, the end performance advantage of the C++ version over the
Python version makes up for that in some specific cases, but not
enough for me to ever want to go back to C++ as a language of first
choice for most of my target development, even with the STL, strings
and other templated classes.

-- David

 

[j_mckitrick]

Of course, the end performance advantage of the C++ version over the
Python version makes up for that in some specific cases, but not
enough for me to ever want to go back to C++ as a language of first
choice for most of my target development, even with the STL, strings
and other templated classes.

I agree. I've been working on an app in Python, next to a rather
large commercial app in MS C++. The error messages I get back from
VC++ are like a maze of tokens and expressions. The Python errors I
get are succinct and to the point. I've been deciphering VC STL
messages for some time now, but they don't really get much easier.
Python makes it downright simple.

jonathon

 

[James Moughan]

--Kamilche

I am no C++ expert, to put it mildly, but couldn't some of your problems
have been solved by using C++ instead of C? C++ has data structures like
vectors, lists, and maps, with many algorithms in the STL to work with them.
Using references to pass arguments, one can avoid low-level pointers in many
cases. C++ strings are easier to use the C char's. There are templated classes
to replicate some of the functionality of Python's Numeric/Numarray or Fortran
90/95 arrays. C is more of a low-level systems language than an application
programming language -- comparing it to Python seems unfair to me, when the
choice of C++ exists.

Heresy, but - often I find C++ a more natural way to express
algorithms than Python. In particular iterators (OK, being blunt -
pointers) are often a great way to break down complexity.

Unfortunately, the amount of junk you have to deal with to get there
is just not worth it.

For strictly numerical tasks Fortran 95 is in my experience both more readable
than Python (because of variable declarations and the ability to define constants)
AND much faster. You do not have to suffer Python's performance hit to program
in a clean, high-level language.

(I like Python, but there is so much pro-Python propaganda in this newsgroup
that some anti-Python messages may be a healthy corrective.)

I keep hearing great things about F95; I'll have to try it out.