how to avoid leading white spaces

[Roy Smith]

I wrote:
>> Another nice thing about regexes (as compared to string methods) is 
>> that they're both portable and serializable.  You can use the same 
>> regex in Perl, Python, Ruby, PHP, etc.

In article <4de9bf50$0$29996$c3e8da3$5496439d at news.astraweb.com>,
 Steven D'Aprano <steve+comp.lang.python at pearwood.info> wrote:
> Regexes are anything but portable. Sure, if you limit yourself to some 
> subset of regex syntax, you might find that many different languages and 
> engines support your regex, but general regexes are not guaranteed to run 
> in multiple engines.

To be sure, if you explore the edges of the regex syntax space, you can 
write non-portable expressions.  You don't even have to get very far out 
to the edge.  But, as you say, if you limit yourself to a subset, you 
can write portable ones.  I have a high level of confidence that I can 
execute:

^foo/bar

on any regex engine in the world and have it match the same thing that

my_string.startswith('foo/bar')

does.  The fact that not all regexes are portable doesn't negate the 
fact that many are portable and that this is useful in real life.

> > You can transmit them over a network
> > connection to a cooperating process.  You can store them in a database
> > or a config file, or allow users to enter them on the fly.
> 
> Sure, but if those sorts of things are important to you, there's no 
> reason why you can't create your own string-processing language. Apart 
> from the time and effort required :)

The time and effort required to write (and debug, and document) the 
language is just part of it.  The bigger part is that you've now got to 
teach this new language to all your users (i.e. another barrier to 
adoption of your system).

For example, I'm working with MongoDB on my current project.  It 
supports regex matching.  Pretty much everything I need to know is 
documented by the Mongo folks saying, "MongoDB uses PCRE for regular 
expressions" (with a link to the PCRE man page).  This lets me leverage 
my existing knowledge of regexes to perform sophisticated queries 
immediately.  Had they invented their own string processing language, I 
would have to invest time to learn that.

As another example, a project I used to work on was very much into NIH 
(Not Invented Here).  They wrote their own pattern matching language, 
loosely based on snobol.  Customers signed up for three-day classes to 
come learn this language so they could use the product.  Ugh.