Lisp development with macros faster than Python development?..

[seberino]

I've been reading the beloved Paul Graham's "Hackers and Painters".
He claims he developed a web app at light speed using Lisp and lots
of macros.

It got me curious if Lisp
is inherently faster to develop complex apps in. It would seem if you
could create your own language in Lisp using macros that that would be
quite an advantage....

I realize that Python has operator overloading and OOP so I'm not sure.

Any ideas? Any *evidence* one way or another?

thanks!

Chris

 

[Benjamin Niemann]

I've been reading the beloved Paul Graham's "Hackers and Painters".
He claims he developed a web app at light speed using Lisp and lots
of macros.

It got me curious if Lisp
is inherently faster to develop complex apps in. It would seem if you
could create your own language in Lisp using macros that that would be
quite an advantage....

I realize that Python has operator overloading and OOP so I'm not sure.

Any ideas? Any *evidence* one way or another?

If that means that I have to learn a new programming language for every
program I'd like to apply a minor fix/customization to: not a good idea.
Probably not just for the casual hacker like me, but also for the
maintainance phase of a project, when the former lead gurus lost interest
and/or employment.

 

[Joost Jacob]

Very hard to say.

LISP has OOP too, Google for CLOS.

Operator overloading is something to avoid anyway, IMHO, just like
static typing is something to avoid if you need fast development, on
schedule and the like.

LISP has one thing that Python does not have: LISP code is LISP data.
A thorough study comparing LISP and Python in this respect would be
welcome.

 

[Dave Brueck]

I've been reading the beloved Paul Graham's "Hackers and Painters".
He claims he developed a web app at light speed using Lisp and lots
of macros.

It got me curious if Lisp
is inherently faster to develop complex apps in. It would seem if you
could create your own language in Lisp using macros that that would be
quite an advantage....

I realize that Python has operator overloading and OOP so I'm not sure.

Any ideas? Any *evidence* one way or another?

Well, his Viaweb company was founded in about '95, right? So he probably just
used Lisp because Python wasn't as well known yet. ;-)

IMO one of Python's strengths over Lisp is that it plays well with many other
technologies. It should be remembered that Graham's use of Lisp in Viaweb was in
building a web application, so that Lisp's main links to the "outside world"
were the filesystem and Apache. They didn't use a database, and Apache had to be
modified in order to work with Lisp.

 

[seberino]

Well, his Viaweb company was founded in about '95, right? So he probably just
used Lisp because Python wasn't as well known yet. ;-)

David

That is what I thought too. It makes sense but I wasn't sure. Still
ain't.
The problem is that questions like 'What lang is fastest to develop
in?'
are hard to answer definitively.

Chris

 

[Peter Hansen]

The problem is that questions like 'What lang is fastest to develop
in?' are hard to answer definitively.

That's because the answer depends on lots of context such what is the
problem domain and who is the programmer. Really, it's an impossible
question to answer. Measuring two different programmers with two
different languages would give useless results, since you can't ensure
the two programmers are equally capable. Using the same programmer for
the same problem with two different languages doesn't work since she
would learn from doing it the first time and inevitably have advantages
in the second time through. Using two different problems is of course
pointless as well.

Ultimately, however, the question needs no answer, since "speed of
development" is not the only or even the most important factor in
development. Take into account issues of maintainability, quality of
code, performance and such, and you've muddied the question even more,
making this a religious question and one that has no meaningful answer,
nor a need to be answered in a simple way.

For *me*, Python has proven to be the fastest language to develop in, so
far, but it's quite possible to imagine another language in the future
which I might learn which would be -- for me -- even faster, at least
for certain types of problems. (But so what? )

-Peter

 

[Kirk Job Sluder]

I've been reading the beloved Paul Graham's "Hackers and Painters".
He claims he developed a web app at light speed using Lisp and lots
of macros.

It got me curious if Lisp
is inherently faster to develop complex apps in. It would seem if you
could create your own language in Lisp using macros that that would be
quite an advantage....

Well, for me, I was playing around with lisp and developing in python
for my little scripts until I finally hit something that could leverage
a lisp macro to create a new type of iterator. After that things moved
foward at a rapid speed. Another point in lisp's favor is that I prefer
nested s-expressions to the sometimes ugly mashup of functional and OO
calls that I end up creating in python.

On the other hand, python benefits from a much richer standard library,
so some of the things you take for granted in python such as
string.split(None,3) need to be found or created.

 

[Mike Meyer]

David

That is what I thought too. It makes sense but I wasn't sure. Still
ain't.
The problem is that questions like 'What lang is fastest to develop
in?'
are hard to answer definitively.

True. You might start by asking which lets you write the fewest LOC,
as studies during the 70s showed that programmers tended to write the
same number of LOC/day, regardless of the language chosen.

The problem with the LOC measurement is that the most productive days
are the ones where you refactor and eliminate a thousand LOC. That
sort of throws the whole thing off.

<mike

 

[Raymond Hettinger]

I've been reading the beloved Paul Graham's "Hackers and Painters".
He claims he developed a web app at light speed using Lisp and lots
of macros.

It got me curious if Lisp
is inherently faster to develop complex apps in.

With Lisp or Forth, a master programmer has unlimited power and
expressiveness. With Python, even a regular guy can reach for the
stars.


Raymond

 

[Raymond Hettinger]

The problem is that questions like 'What lang is fastest to develop
in?'
are hard to answer definitively.

FWIW, Google's answer to that question is C++, Java, and Python. For
any given problem, any of the three are acceptable. Each programmer or
engineering team gets to decide based on his or her language
expertise.*



Raymond


* Source: Greg Stein's keynote address at PyCon 2005.

 

[Fuzzyman]

So Lisp is for really good programmers, and Python is for mediocre
programmers ?

Best Regards,

Fuzzy
http://www.voidspace.org.uk/python

 

[Michele Simionato]

Fuzzyman:

So Lisp is for really good programmers, and Python is for
mediocre programmers ?

Python is *also* for mediocre programmers. I see this as a
strength, not as a weakness.

Michele Simionato

 

[Andrew Durdin]

With Lisp or Forth, a master programmer has unlimited power and
expressiveness. With Python, even a regular guy can reach for the
stars.

+1 QOTW

 

[Fuzzyman]

Fair enough ;-)

I'd like to discover the power of Lisp, but I have a limited amount of
time to sink into programming... so maybe I'm better off putting my
energies and imagination into Python.

*A language is a medium of expression.* - Paul Graham

All the best.

Fuzzy
http://www.voidspace.org.uk/python

 

[Antoon Pardon]

Fuzzyman:


Python is *also* for mediocre programmers. I see this as a
strength, not as a weakness.

But sometimes I get the impression people want it to evolve
so it is only for mediocre programmers.

 

[Larry Bates]

You don't say how long it took to develop the "macros" but
you should see what kind of website an experienced Zope/Plone
programmer can whip up in a few minutes.

Acceleration in programming has always been about the "Standard
Library" (not only Python's standard library but also your
standard library). I'm talking about stable, debugged,
documented macros, functions and classes that the programmer can
use to quickly do very complex tasks. I discovered this well
over 30 years ago and taught many young programmers and
University students this very important "trick". As you write
software you will eventually come across common routines that
are used in almost every software package (logging, reading
from CSV files, date/time manipulation, ...). If you have lots
of these to choose from in your library, you will produce code
10-100 times faster than those that start over every time AND
the code will be MANY times more reliable because you are
utilizing stable code that has been debugged over a long period
of time. You will also be producing code that is more
"maintainable". Discovered a bug in a library routine? Fix it
and then it is fixed in 100's (or 1000's) of existing programs
that use it.

-Larry Bates

 

[Zachery Bir]

You don't say how long it took to develop the "macros" but
you should see what kind of website an experienced Zope/Plone
programmer can whip up in a few minutes.

Zope/Plone (as frameworks) represent exactly the kinds of DSLs people
have been building with Lisp for decades. Shoulders of giants, and all
that. Lisp has web app frameworks as well: Uncommon Web and BKNR, to
name two.

Zac

 

[Tom Anderson]

True. You might start by asking which lets you write the fewest LOC, as
studies during the 70s showed that programmers tended to write the same
number of LOC/day, regardless of the language chosen.

The problem with the LOC measurement is that the most productive days
are the ones where you refactor and eliminate a thousand LOC. That sort
of throws the whole thing off.

As in:

http://www.folklore.org/StoryView.py?project=Macintosh&story=Negative_2000_Lines_Of_Code.txt

Perhaps the real question, then, is which language allows you to delete
lines of code most quickly.

tom

 

[Peter Hansen]

Perhaps the real question, then, is which language allows you to delete
lines of code most quickly.

No, then the question becomes "which language allows you to quickly
write very many lines of code which then have to be deleted".

Of course, writing those lines manually would be silly if you could
automate the process. After all, the lines of code aren't required, so
they don't really have to do anything, do they?

So naturally Assembly would be the proper way to get maximum performance
out of your automatic code writing program.

Therefore Assembly is clearly the fastest development language.

-Peter

 

[Rocco Moretti]

FWIW, Google's answer to that question is C++, Java, and Python. For
any given problem, any of the three are acceptable. Each programmer or
engineering team gets to decide based on his or her language
expertise.*

Actually, Google's answer to that question is something called "ILOG
CPLEX", followed by Visual Basic, English as a second language, PHP, and
"Holt Software Associates". ;-)

http://www.google.com/search?hl=en&q=What+language+is+fastest+to+develop+in?&btnG=Google+Search

Given this finding, I'm not sure I should put much weight into Google
search results anymore ...