I'm using cygwin, which includes octave (matlab work-alike), postgresql, and
perl. I also have Excel 2003. I'll be analyzing data from the I/O log files of
a suite of applications that communicate in a "dataflow" manner, by which I mean
that the data flows unidirectionally, though in parallel paths at some points.
Different format log files, different delays along the path. I'll be studying
things like propagation times and dropped messages. I expect to cross reference
log entries from different log files. Which language would be best?

I have used c++/STL quite a bit, but it may be a bit low level.

I've tried Perl, but have always been perplexed by the syntax, though I'm OK
with unix and regular expressions.

I'm well versed with matlab, which I could use Octave for.

I can fumble around in Excel.

I've only read about postgresql, though I used Oracles SQL for a few months, but
this was decades ago. I'd be picking it almost from scratch, but not quite.

I anticipate the parsing of the various log files into databases or data
structures to be a grueling task.

What is a good strategy in terms of the above options for the task? Though I
expect that time invested in the chosen strategy will pay off in future
analyses, the current effort is not a long term effort in the sense that I will
have luxury of slowly getting to know a more powerful but more cryptic language.

Thanks for any thoughts on this.

Dubious Dude wrote:

As of 22/11/2007, that would be Python.

Hello,
At the begging computer ware slow, today machines are fast. Use
their power - choose solution which allows you to concentrate on real
problems. Perl is good choice, but... I personally prefer Python. I
have heard that Perl code before and after MD5 - look the same <g>.
Please try Python, Ruby. Python and Ruby are modern scripting
solutions with big knowledge base, dedicated code libraries, large
communities. All of them are open and growing fast (libraries,
communities, knowledge bases...). If your plans are long lasting -
choose solution that will be popular in the future. Now Java/.Net are
popular - but their application servers are rather complicated to play
with... I'm afraid - business is not ready to use them right now, so
in short term learning Python / Ruby is not so profitable as learning
Java or C#.

--
Regards,
Micha³ Zaborowski (TeXXaS)

On Nov 22, 6:51 am, Marco Mariani <ma... (AT) sferacarta (DOT) com> wrote:
I've dabbled in Python tutorials. The fact that indentation space
affects the meaning of the code is something I haven't gotten use to.
Regular expressions mechanisms seem to be more verbose than I'm use to
(sed, vim, grep), but then again, so it is the case for Perl, though
less so.

On Nov 22, 7:02 am, michal.zaborow... (AT) gmail (DOT) com wrote:
Yes, though I think it is possible to exercise the discipline to write
good code.

Mister.Fred.Ma (AT) gmail (DOT) com wrote:

The only time it's been a problem for me in 8 years is when somebody
kept using hard tabs instead of spaces.

Otherwise, even new python programmers usually see the indentation as a
big plus of the language, for several reasons (readability, vertical
space savings, etc).

Coming from perl, I've seen that regular expressions are much less used
in python. Sometimes avoiding them (and using iterators, dictionaries,
whatever) even results in faster programs.