[Tutor] a quick Q: how to use for loop to read a series of files with .doc end

[Alan Gauld]

On 08/10/11 14:17, lina wrote:

>     You will never say, "Gosh, I wish I knew FEWER programming languages!"
> Ha Ha ...
>
> regarding the python and C, I was told that python is good for interface
> and others, but slow. and further was suggested to write the main part
> in C and wrapped in python.

The myth that Python (and interpreted languages in general) are too slow 
for "real programming" is common among older programmers(*) who were 
trained on old computers like the original PC and servers like the DEC 
PDP series. When I left university our office had just acquired the 
largest, most powerful VAX computer in Europe - it ran at 12MHz and had 
16MB of RAM and supported over 100 users! Wow! However many. many, 
applications can now be written in interpreted languages and run fast 
enough (and the key word is "enough").

If you really need the last bit of speed then yes, C (or even assembler) 
may be needed, but that is a tiny proportion of the code for most 
things. So to take your friends comment and turn it around:

"... write the main part in Python (or .NET or Java or...) and write the 
speed critical parts in C."

Because typically the Python part is likely to be 70-90% of the code
and the C part much less.

(*)I tend to consider myself in that group and up until about 15 years 
ago held the same view. But around then I got involved in projects using 
languages like Smalltalk, Tcl/Tk and VB. It was obvious that these 
languages were perfectly capable of producing usable apps
on the typical hardware of the time - early Pentiums running at
around 500MHz and with 128M RAM etc. Nowadays with most PCs having 
processors running in excess of 2GHz and over 2GB RAM there are very few 
apps that cannot be constructed using interpreted languages, and many 
are. I suspect the reason most commercial projects still use Java, C++ 
etc is twofold:
1) Magazines etc tend to review apps against their competition on speed, 
even if the real world speed measures are irrelevant. So for marketing 
purposes you must be seen to be comparable to your competitors.
2) Compiled code is harder for the competition to reverse engineer so 
coding secrets are 'safer'. (This is largely psychological since its the 
design concepts your competitors are more likely to steal, not the 
algorithmic details.)

There are projects where ultimate speed is still important, indeed one 
of my current projects at work is being largely rewritten from C#.Net to 
compiled C++ for exactly that reason. (But it is a number crunching 
server app supporting around 15000 concurrent users) But the vast bulk 
of user programs could be written in interpreted languages IMHO.  One of 
the interesting things about my recent conversion from Windows to Linux 
for my home PC  is to see just how many of the desktop apps are written 
in Python, including games and office apps etc.

-- 
Alan G
Author of the Learn to Program web site
http://www.alan-g.me.uk/