in reply to Re (tilly) 2: Technical Interview
in thread Technical Interview

Well, I only used the &foo syntax for clarity of expression: "&foo" is the one and only name of the function "foo", while foo() could refer to an operator. When teaching or explaining about function calls I think it reasonable to use the ampersand. In real code, of course, I avoid ampersands whenever possible.

But to answer your question: I'd probably smile broadly and hire the candidate on the spot. If there was no hurry, I would first indulge in a thoroughly enjoyable argument over how attempts to improve readability can break things by accident. (Contrary to popular folklore, &func is not synonymous with func(@_); the former shares a single @_ array, while the latter makes a temp copy that disappears when func returns. Don't feel bad if you didn't know that; I actually caught TheDamian in the same mistake at TPC5.)

It's certainly true that a good programmer is a good programmer no matter his background. But background can take time to overcome, and some habits are difficult to unlearn. A really good programmer must grok his tools, above all his language. I was hired in my current job to mostly code Python (ack! spit!); I thought for sure I'd have no problem with it, but it's taken a few months to get get to the point where coding in Python isn't like translating English to Spanish word for word. (Which, incidentally, sucks.)

UPDATE: Fixed usage of the word "latter", per tilly's kind correction.

    -- Chip Salzenberg, Free-Floating Agent of Chaos

Replies are listed 'Best First'.
Re: Re (chip): Re3: Technical Interview
by TheDamian (Vicar) on Dec 31, 2001 at 03:10 UTC
    Don't feel bad if you didn't know that; I actually caught TheDamian in the same mistake at TPC5.
    Well, it's certainly true: chip did indeed catch me out at TPC5. And in front of 400-500 people!

    Though in fairness, my mistake occurred in SelfGOL: quite possibly the most complex 1000 bytes of Perl ever written.

    But it wasn'tbecause I was unaware of the @_-sharing nature of &foo. In fact, the code in question deliberately abused that feature. I messed up because I (carelessly) relied on the behaviour once to often within the same block.

    Nevertheless, I was very grateful to chip. His correction encoraged me to make SelfGOL even more diabolical. As indeed has this thread.

    >;-)

[reply]
[d/l]
[select]
      Damian is right about SelfGol; it is excruciatingly diabolical. It evidences a skill in Perl compression and obfuscation far beyond my own. My work with Perl has mostly been under the hood, so to speak. Knowing your car can help you be a good driver, but it won't turn you into Evel Knievel. Damian's coding stunts left his TPC5 audience, myself included, in awe.

      But more basic forces were also at work that day. As every teacher learns, nothing quite motivates a roomful of students to cleverness like the thrill of catching the teacher in a mistake.

      Come to think of it, Damian mentioned early in that very presentation that he had recently enjoyed catching Jeffrey Friedl--regex guru nonpareil--in a regex error. Did Damian's own anecdote plant a seed in my mind, which then bore fruit when I noticed his error? Who can know....

          -- Chip Salzenberg, Free-Floating Agent of Chaos

[reply]
        Come to think of it, Damian mentioned early in that very presentation that he had recently enjoyed catching Jeffrey Friedl--regex guru nonpareil--in a regex error.
        Actually, it was the younger regex Jedi, our very own japhy, whom I caught out.
        As every teacher learns, nothing quite motivates a roomful of students to cleverness like the thrill of catching the teacher in a mistake.
        So true. In fact, it was precisely by manipulating that basic human desire (i.e. to see the mighty stumble ;-) that I caught out someone as highly clueful as japhy. I deliberately made it look like I'd made a mistake, when in fact I was demonstrating an inconsistency in a rarely-used behaviour.

        Specifically, I showed some code equivalent to the following: 

            $var =~ s?pattern?string?;

        [download]
        and asked what was unusual about the substitution.

        Most regex adepts will reply:

        The use of ? as the delimiter causes the regex to match at most once between calls to reset (as described in the perlop manpage).

        But the real answer is:

        The use of ? as the delimiter doesn't cause the regex to match at most once between calls to reset. That behaviour only applies to matches, not substitutions.
        I have to confess that I much prefer this proxied variation on chip's "humble the guru" motivational technique, rather than the version in which the presenter himself is the one who gets humbled! ;-)
[reply]
[d/l]
[select]
Re (tilly) 4: Technical Interview
by tilly (Archbishop) on Jan 01, 2002 at 07:58 UTC
    I actually did know that, as is evidenced by the fact that I knew without looking at documentation that the right correction to what you wrote is s/latter/former/. :-)

    (ie &foo shares @_, foo(@_) does not.)

    Those who didn't know this should check out perlsub, or if you don't want to think carefully about consequences, the excellent summary at (tye)Re: A question of style. (Hrm, glancing at it a current version of perlsub gets it right twice, but the comment in the code sample is misleading.)

[reply]

In Section Seekers of Perl Wisdom