Interview: Mark "Markey" Kretschmann (Not the Gentoo Weekly News) [LWN.net]

Interview: Mark "Markey" Kretschmann (Not the Gentoo Weekly News)

Posted Feb 9, 2008 14:42 UTC (Sat) by lmb (subscriber, #39048)
Parent article: Interview: Mark "Markey" Kretschmann (Not the Gentoo Weekly News)

Whenever I see a seasoned programmer not merely express a preference for a language (which is
natural to have, I guess), but a strongly biased opinion and dissing other languages, I know
I've meet someone who still is naive enough to believe in silver bullets.

The cases where the language chosen makes a really significant difference to programmer
productivity are exceptionally rare. Sure, some languages fit certain problem domains better,
or have better libraries for given tasks, or you've got this big existing project to interface
with.

But I'dn't trust a "seasoned" programmer who got hung up about that. A seasoned programmer
should be able to program in all of the major languages within at most a few weeks to get upto
full speed.

That said, everybody knows that the Gods in their infinite wisdom wrote the universe in Perl,
and that God is anxiously waiting on Perl6 so he can end all evil and banish the snake and
mammon's tempting gems, so we shall be back in paradise.

(Log in to post comments)

Amarok is C++

Posted Feb 9, 2008 16:42 UTC (Sat) by eean (guest, #50420) [Link]

You know that Amarok is written in C++ right? Mark obviously doesn't believe Ruby is a silver
bullet.

But thanks for not taking this seriously at least. ;)

The universe

Posted Feb 9, 2008 17:29 UTC (Sat) by rfunk (subscriber, #4054) [Link]

http://xkcd.com/312/
http://xkcd.com/224/

On the other hand....  http://xkcd.com/353/

Interview: Mark "Markey" Kretschmann (Not the Gentoo Weekly News)

Posted Feb 9, 2008 20:32 UTC (Sat) by TxtEdMacs (subscriber, #5983) [Link]

I would not be so quick to hope to banish the snake from the world.  In some versions of text
from which the old testament was selected; it was the snake that spoke the truth.  Hence, the
credo that no good deed goes unpunished seems to be the more valid lesson.  Therefore, I would
suggest less religiosity, let's base our arguments on observable facts.  That is, even when
they are our deeply held preferences.  Remember, we are talking about a scripting language not
a panacea to the World's ills.

Interview: Mark "Markey" Kretschmann (Not the Gentoo Weekly News)

Posted Feb 12, 2008 15:18 UTC (Tue) by ikm (subscriber, #493) [Link]

> I would not be so quick to hope to banish the snake from the world; it was the snake that
spoke the truth

Keep the python.

Language productivity

Posted Feb 10, 2008 18:41 UTC (Sun) by man_ls (subscriber, #15091) [Link]

But it is a proven truth that, while programmer productivity measured in lines of code is more or less constant in any language, the functionality provided does indeed change a lot depending on the language. It takes more lines to write something in low-level languages than in high-level languages. That is why high-level languages exist, and why assembler is only used rarely nowadays.

If you want hard numbers, data has been tabulated from function point statistics. Now, function points may not be as precise as they are often sold, and these measures vary a lot; but they definitely give some perspective. Even if a good programmer is proficient in any language in a few weeks, writing web applications in low-level languages or writing graphic applications in assembler is madness. And just try to write a hello-world in Brainfuck.

Now, all of these languages (Ruby, Python, Perl) are in a similar league, so it may be hard to see the differences; but your blanket statement:

The cases where the language chosen makes a really significant difference to programmer productivity are exceptionally rare.

gives the wrong impression, even if you probably know better. Language does make a huge difference in productivity and it should be chosen carefully.

Language productivity

Posted Feb 10, 2008 21:15 UTC (Sun) by i3839 (subscriber, #31386) [Link]

You may be right if you only consider the time coding something. But the time spend writing
code is in general only a small part of the whole thing. And that SLOC/FP count thing is
silly, because it doesn't consider how much time it takes to write those lines. Many lines of
simple code is often better than less, more complex lines of code, and takes less time
writing.

When writing something like a web application, it doesn't really matter what language you use,
what's important is what framework you use. And that's true for most things, existing
libraries and frameworks are the ones reducing the amount of work most, not the language
choice itself. So the huge difference comes from there. I don't say that language choice
doesn't matter, it does, but it's much less than the impact of the included standard libraries
and easily available libraries.

I wonder if there are Python bindings for C...

Language productivity

Posted Feb 10, 2008 21:40 UTC (Sun) by man_ls (subscriber, #15091) [Link]

I don't say that language choice doesn't matter, it does, but it's much less than the impact of the included standard libraries and easily available libraries.

In that you are completely right. Witness Arc's challenge: the inventor of this new language, Paul Graham, is challenging people to come up with a shorter version of a barebones web application. And the responses: of course the framework is what counts, not the language.